User:Glrx/sandbox

From Wikipedia, the free encyclopedia

Unicode NFC and article titles[edit]

  • Should Horatio Nelſon be blue and link to Horatio Nelson?
    • Nelson Nel%C5%BFon Nel%25%26%23x17F%3Bon
  • Führer (composed) versus Führer (decomposed); apparently not a redirect; the links were normalized.
    • F%C3%BChrer Fu%EF%BF%BD%3Fhrer Fu%26%23x308%3Bhrer

Flag Books[edit]

Power supply voltages[edit]

Analog

  • ±15 (early transistor and opamp; 40V transistors)
  • ±12
  • ±5
  • +5

Logic

  • ±?? ±?? (1401)
  • +3.6 (integrated RTL; I think it was 3.6V; funded by satellite needs and power supply = 3 NiCd)
  • CosMOS (RC) 3-15V
  • +5 TTL; 0.8 to 2.8V; pullup not strong;
  • -5.2V MECL. Negative supply so NPN transistors and ground ref. Why 5.2 and not 5.0?
  • 3.3V CMOS discrete; any bipolar?; goal to lower power consumption; keep 2.8V? 5*.707 → 3.535V.
  • 2.0V
  • 1.8V
  • 1.5V
  • 1.2V
  • 1.0V

Significant transistors[edit]

  • CK722 Raytheon. Ge PNP. One of the first plastic?
  • 2N696 and 2N697 Si NPN mesa; no epi; before thick oxide known as a diffusion shield. BTL made silicon; somebody had Ge mesa. I think mesa etch was to reduce C_bc. Did etch have an advantage with exposed base–collector junction? Why not make smaller die? I think PIN and Snap diodes want uniform field, so they have junctions exposed by dicing.
  • 2N1304 and 2N1306 Ge NPN/PNP. Typical switch. Which transistor types used on 1401 (033 / 034)? IBM document had a list.
  • what was the typical Ge power transistor? 2N176?
  • 2N2222 Si NPN epi planar annular
  • 2N3055 Si NPN power. Original was hometaxial, but later copies went to epitaxial. Hometaxial is single diffusion: https://encyclopedia2.thefreedictionary.com/hometaxial-base+transistor OMG, http://powersupply.blogs.keysight.com/2012/09/early-power-transistor-evolution-part-2.html describes having the wafer be the base, simultaneous diffusion of collector on one side and emitter on the other, then an emitter-side mesa etch down to the base. With the base exposed, a base connection could be made. Why fancy emitter geometry?
  • 2N3904 and 2N3906 Si NPN/PNP.
  • 2N5109 and 2N5160

RCA Power Transistor Applications Manual (1983) has descriptions hometaxial, etc.

Cross-evaporation: US 3709695, Bowman, R., "Fabrication of Semiconductor Devices", issued 1973-01-09, assigned to Motorola 

Costas loop[edit]

The Costas loop article needs work. See Talk:Costas loop.

The angle addition formulas:[1]

We can use those formulas to find the result of a phase shift.

Angle addition formulas taken together produce the heterodyne formulas:

These clarify the first image in the article that has a simple story of in phase and quadrature. The diagram uses a sin() phase carrier and a sin() phase VCO.

  • Costas loop model.svg
    Costas loop model.svg

The image does not give the heterodyne formulas, but the image shows a zero phase difference between the carrier and VCO because both are described as . To model the phase difference, let a be the carrier phase and b the VCO phase. The upper and lower branches then produce:

The summed phases get filtered out. Demodulation is on the upper branch: cos(a-b) is approximately cos(0) = 1. Phase control on the lower branch. The lock point is stable at near zero phase difference: if b gets a little ahead of a, then sin(a-b) is negative which will reduce the VCO phase.

This diagram started the article, has the cleanest math, and the rest of the article should follow suit. Top branch should be demodulation and bottom branch phase control. The phases going to the mixers should be sin() and cos().

In contrast, the two diagrams in the final section of the article are confused, inconsistent, and changing.

  • Costas loop before sync.svg
    Costas loop before sync.svg
  • Costas loop after sync.svg
    Costas loop after sync.svg

The recent formulas given in the diagrams for the lower branches are:

We should get the lower formula by substituting for , but that gives

Notice the -θs; that makes the lock point stable for small values. This last formula is identical to the cos case.

In other words, the formulas says the first diagram locks with a phase difference near 0, and the second diagram locks with a phase near 180°.

If we look at earlier versions of the two diagrams, the situation is still confused. The "before sync" SVG file from 2012 recognizes that the -90° phase shifter creates a minus cosine, so it negates m(t). That makes the loop unstable for small phase differences between the carrier and the VCO. The "after sync" SVG from 2012 is even more confused. It assumes the phase difference between the VCO's sin() and the carrier's sin() is zero. That is evidenced by the sin(0) expression in the lower branch. It expects the top branch to provide the demodulated m(t), so it is trying to mimic the diagram in the first section. It does not show the phase error, so we cannot comment on stability.

The same issues are present in another diagram:

  • Costa loops blocks2.svg
    Costa loops blocks2.svg

The expressions for the upper and lower branches with a −90° phase shift to the lower branch should be (with a the carrier phase and b the VCO phase):

And -sin(-x) = sin(x) by antisymmetry, so the "after lock" lower branch formula is correct, but the lock is unstable as shown. IF the VCO is a little fast, then the phase difference with be positive, and the feedback will speed up the VCO even more. The diagram is wrong (assuming positive filter gain).

Amelia Earhart[edit]

Radio equipment[edit]

Amelia Earhart#Radio equipment

Is a confusing mess.

Cyril D. Remmlein is in Bendix Radio Research Division. In February 1937, Remmlein brought a prototype Bendix Receiver to Newark, NJ airport. The first receiver was still in manufacturing. O. Vernon Moore, the design engineer, had not designed the coils yet. Earhart and Manning choose 200 to 10,000 kHz in 5 bands. They would get the first production receiver; it would be shipped to Burbank when ready.[1]

Joseph Gurr is UA radio tech at Burbank. Comes on the scene because WeCo radio did not work. Gurr quickly discovers antenna not plugged in. "From then on, Gurr was the person they called whenever they needed a radioman."[2] Become go-to radio guy for Earhart and Putnam. Gurr had nice comments: you must do the last hundred miles on radio. https://tighar.org/wiki/Modifications_by_Joe_Gurr

Gillespie says plane shipped with starboard ventral antenna and a trailing wire antenna from the tail. https://tighar.org/wiki/NR16020_antennas#Chronology Pitot tube to two ventral masts. https://tighar.org/Publications/TTracks/1995Vol_11/starborn.pdf says TA is just below tail light on delivery; also says not the TA that was removed; comment at end of article about pinning down nomenclature of the loop.

October 1936 has installation of Hooven.

Jan-Feb 1937 had installation of Vee. (I think this was a Bell Labs installation; they could reasonably install the WeCo receivers.)

However, probably installation of Bendix RA-1 receiver. Who is Remmlein? He flew out from Washington DC to do the modification.

Gas tanks installed (limiting access to tail). Also during that time, trailing wire antenna moved to ventral mast underneath cabin.

Somewhere, the trailing wire has a weight. That would incease vertical component of antenna and improve forward transmission. Long & Long (1999, p. 173): "Also installed was a 250-foot trailing wire antenna, which was weighted on the end and wound onto an electrically driven reel in the Electra's tail. Low frequencies, such as the 500=kilocycle channel, require long antennas and communicate in code only." The "in code only" is dubious; lower end of BC band is 550 kHz.

Just before flight, Bendix donates $2,500.

See 1941 Bendix Brochure for MN-31 Automatic Radio Compass.

Hooven removed in early March 1937. Gillespie says Hooven Dome replaced with Bendix MN-5 loop. Long and Long say the loop is a Navy RDF-1-B loop with a custom coupling box.[3]

Dorsal antenna[edit]

TIGHAR probably has it right that dorsal was transmit-only and the transmitter lead went from transmitter to leg of vee.

Source: 1962 letter from W. C. Tinus, Vice President of Bell Telephone Laboratories:[4]

I was the radio engineer who was responsible for the design and installation of her radio communications equipment....
I had been a radio operator aboard ship in my younger days and knew the importance of being able to communicate at 500 kc over the oceans. I persuaded Miss Earhart and Mr. Putnam on this point and modified a standard three-channel Western Electric equipment of the type then being used by the airlines to provide one channel at 500 kc and the other two at around 3000 and 6000 kc... A simple modification also enabled transmission to be made on CW or MCW, as well as voice, and a telegraph key was provided which could be plugged in, in addition to a microphone for voice communication. It was my thought that many ships throughout the world had 500 kc radio compasses and could probably better obtain bearings if the key were held down for an extended period while radiating modulated CW (MCW).

There was a reference on design of dorsal antennas. Expect Tinus did a good job; does TIGHAR have the dimensions? Then Gurr changed the length (and the tuning). Didn't work well on trip to Miami, so Miami changed it.

There's an issue of how Gurr switched the transmitter to the 500 kHz range, but that becomes irrelevant if Miami removed the network. Expect a direct connection with no additional loading/matching networks.

Ventral antenna[edit]

TIGHAR probably right here that just went to the receiver.

Could be used for sense antenna, but then lead would have to go up to the RDF unit.

TIGHAR wipe-off theory is probably right, too. Earhart's receiver still worked because she got the RDF receiver to work Howland. That means the fuse probably did not blow. Earhart could not get the 3 or 6 kHz channels to work, so the antenna was probably disconnected or wiped off. Alternatively, the controls may not be set properly, but Earhart had that working at Darwin.

There are more complicated theories about connection....

Earhart RDF[edit]

There are pictures of Earhart with the loop and coupling box. https://tighar.org/smf/index.php?topic=770.15 has pictures. Box has controls

  • "BAND" selector: 1, 2, 3, 4, 5
  • function selector: possibly B, D, R (these initials were standard by 1936 with RDF-1-A's L1130-B coupler; see Aircraft Manual)
  • "LOOP TUNING"
  • "A" antenna binding post
  • "ON"/"OFF" switch
  • three other features that are not clear

Long & Long (1999, p. 63):

The loop coupler control box had been custom-made to electronically adapt the standard Navy RDF-1-B direction-finder loop to the new Bendix RA-1 receiver. Five frequency-band numbers were shown on the face of Earhart's loop coupler, but the Navy model had six bands. The actual frequency range of each band was listed on a chart concealed inside the inner bottom cover plate of the unit. The limited space on the face of Earhart's loop coupler required that the frequency bands be labeled 1, 2, 3, 4, and 5. Remmlein cautioned Manning that the radio direction-finder loop was designed to take bearings only on frequencies between 200 and 1430 kilocycles, whereas the receiver could receive signals on frequencies up to 10,000 kilocycles. If Earhart or Manning tried to take a bearing on a high-frequency signal using the Bendix receiver's upper bands, the resonance of the loop would be so far out of tune they probably could not get a minimum signal.

Next paragraph says 5 bands do not match.

Notes for page 63 cite to the Bendix RDF-2-A Navy instruction book, page 12, for "limited space". A 21 May 1976 interview of Al Hemphill, Bendix direction finder engineer, "If Earhart or Manning tried" the loop would be so far out of tune they could not get a minimum, and 20 January 1976 interview of Joe Gurr for "natural mistake" of matching band numbers.

RDF was useless flying into Africa. https://tighar.org/Projects/Earhart/Archives/Forum/Forum_Archives/199910.txt (apparently from Long and Long quoting letter from Noonan dated 6/8/1937.

Gillespie dubious of Long: "From what I've seen, Elgen presents no documentation to support most of his allegations about the Bendix DF and he completely neglects to mention the earlier Hooven/Bendix Radio Compass that was installed and later removed." (Forum Archives 199910.txt)

IIRC, the notes section of the book refers to the RDF manual (a later version that I did not find).

  • US 2144309, Hicks, Charles William, "Radio Apparatus", published 12 July 1935, issued 17 January 1939, assigned to Bendix Radio 
RDF-1A / L1130-B[edit]

See description below.

R T D B

RDF-1B see http://aafradio.org/docs/Navy-radio-gear-1943.pdf

RDF-2-A / L1678A[edit]

Bendix documents (Reference Drawing AR3000) definitively identify Earhart's RDF as an RDF-2-A with an L1678A coupler, but the units may be prototypes.

DU-1 RDF[edit]

There's a picture of a later coupler that has operating instructions.[2] Much better picture at National Air and Space Museum: https://airandspace.si.edu/multimedia-gallery/4523hjpg Radio Compass, Navy, DU-1, Bendix. The loop is two-tone color.

  • R RECEIVE ONLY — DO NOT ATTEMPT TO TAKE BEARINGS
  • B TUNING — TUNE LOOP CAREFULLY FOR MAX SIGNAL
  • D DIRECTION — SWITCH TO D AND SWING LOOP FOR MAX SIGNAL — ARROW WILL POINT TO STATION. ADJUST MASK SO THAT ZERO MARK IS OVER POINTER WITH LOOP IN POSITION OF MAX SIGNAL.
  • B BEARING — RETURN SWITCH TO "B?". SWING LOOP TO MIN IN UNMASKED PORTION OF DIAL. BEARING IS RELATIVE TO PLANE HEADING.

R connects input omni antenna (terminal "A") to output (to receiver via J102). RDF unit may be off. B would add capacitors to resonate without adding omni. Omni input grounded; just loop operative. Loop tuning would be a variable capacitor to fine tune the loop. Variable capacitors do not have wide range (e.g., 40 to 360 pf is range of 3 for broadcast band). Loop range of DU-1 is 200 to 1600  kHz. That's 3 octaves. That's why the picture with instructions would have 3 bands. The bands are 200 to 400, 400 to 800, and 800 to 1600. Earhart's coupler may have used smaller range variable capacitor and therefore used 5 bands. D setting includes omni antenna to get a cardioid pattern. There's an on-off switch. DU1 uses two 12SK7 pentodes; power may be off in "R". http://www.r-type.org/pdfs/6sk7.pdf

There's some nice cleverness here. The loop has a fixed 200 pF capacitor (C-201). The tuning variable capacitor (C-102) is 19–395 pF. Autotransformers (T-101, T-102, and T-103) are used to boost the effective capacitance. For example, to get an octave range over the 200 pF fixed capacitor, one would need 0–600 pF. A transformer can lower the capacitors apparent impedance (increase its apparent capacitance). That gets around the limitation of the RDF-1A that switched fixed capacitors into the lower ranges, so the lower bands covered much less than an octave.

Gillespie[edit]

Gillespie quotes

  • "Aero Radio Digest: Newest Developments in the Field of Aircraft Radio". Aero Digest. 30 (??): 42–44. August 1937. ISSN 0096-4344. They may be used as fixed-loop homing devices or as navigational direction finding instruments within frequency range of 200–1500 kcs. {{cite journal}}: |section= ignored (help); not on Internet Archive.
    obtained from library. Bendix announces COMMERCIAL RDFs that MN-1, MN-3, MN-5, MN-7 (but also mentions MN-2!). Units cover 200 kHz to 1500 kHz.

See https://tighar.org/wiki/Radio_equipment_on_NR16020

August 1937 issue of Aero Digest magazine: “Bendix D-Fs are designed to operate in conjunction with Bendix Type RA-1 receiver, but will also give accurate and dependable bearings when used with any standard radio receiver covering the desired frequency range.” The article also notes that these receivers can be used “as navigational direction finding instruments within frequency range of 200–1500 kilocycles.

Also https://tighar.org/smf/index.php?topic=452.120 where he offers a PDF.

Comments about second ventral antenna used as Hooven sense antenna.


  • Comment. The 1941 Bendix catalog does not include the MN-1; instead it has the MN-13 Direction Finder on page 10. That unit covers 150 to 1500 kHz. Blow up picture, and it shows a 3-band unit, so probably similar to the DU-1; DU-1 project was finished by 1941.
    Instruction Book for Model MN-13A Direction Finding Equipment (10 July 1940), copyright entry: https://books.google.com/books?id=gkfQAAAAMAAJ&pg=PA835&lpg=PA835&source=bl&hl=en
  • MN-20 9 inch loop (1941 catalog)
  • MN-24 18 inch loop (1941 catalog)
  • MN-26 (page 11) used the MN-20 loop up to 7000 kHz.
Hefele[edit]
  • US 2251708, Hefele, Edward J., "Direction Finder Antenna System", published 1937-04-27, issued 1941-08-05 
Russian text[edit]

RA-1 receiver[edit]

https://tighar.org/Projects/Earhart/Archives/Research/Bulletins/53_MiamiPhoto/53_MiamiPhoto.htm Shows cockpit control box with an argument about a HF direction finder. But that photo fits with Long and Long description of RA-1 with remote control at copilot's station (sic pilot). Bendix RA-1B was used during the war. Manual: MR-1B remote control. Remote control uses manual speedometer cables for tuning and bandswitching.

Bendix Brochure has RA-1 pictures and MR-1 remote control. http://aafradio.org/docs/Bendix_1941_brochure.pdf

There is no magic RDF in the RA-1 receiver. RA-1B manual shows it as a conventional communications receiver. Its ability for RDF would be detecting minimum and maximum signals. The RA-1B has an antenna selector.

  • FA Fixed antenna. Ordinary reception. Capacitively coupled to band selector through paralleled C81/C82. R47 (500 KΩ bleed) to ground.
  • TA Trailing antenna. C80 in series with paralleled C81/C82. Same antenna input as FA.
  • DF Direction finding. Connects DF antenna input to band selector with no load resistor. FA antenna shorted to ground.

RA-1B copyright entry: https://books.google.com/books?id=kGZbAAAAIAAJ&pg=PA10&lpg=PA10&dq=bendix+manual+for+ra-1+series+receiver+for+aircraft&source=bl&ots=cOiGNkElSE&sig=NwBhD9CV7R7uUO-IsVWpE4ARm1k&hl=en&sa=X&ved=0ahUKEwjXkob_wuXXAhWpsVQKHT4WCNEQ6AEIOzAD#v=onepage&q=bendix%20manual%20for%20ra-1%20series%20receiver%20for%20aircraft&f=false

http://antiqueradios.com/forums/viewtopic.php?f=2&t=274814

Luke Field crash. Suggestion of port side underbelly damage. Did the ground loop damage the port side? Trailing wire and starboard ventral antenna were not reinstalled. There's some background, too. Manning left the flight after the crash.

Manning was the only one who could receive Morse code (15 wpm). Manning was the only one who had an FCC license to transmit Morse (obtained in March 1937). Manning was the only one who could transmit on 500 kHz (Earhart and Noonan were restricted radiotelephone licenses).

Consequently, a 500 kHz transmitting antenna would be surplusage. It may be that the RA-1 was viewed as a Morse accessory, too. That would have it pulled after first attempt but before second. Don't know if sources support that. Gurr did train Earhart on RA-1 in March according to Long and Long 1999.

Further conflict. March was before first attempt. Gurr modified dorsal antenna for 500 kHz. No need to do that unless trailing wire removed.

TIGHAR has this theory that lengthening the Vee created a horrible mismatch. At the same time, there's a comment that the feedpoint changed. Hello, offset dipole. Traps also popular.

Theory that during the Lae takeoff, the belly antenna was lost. Probably caused the loss of the Electra's ordinary receiving antenna.

Sources suggest RDF reception at 7500 kHz is very suspect. Could not tune the loop to that frequency. Loop might receive it, but trying to tune the loop would severely attenuate the signal. Indicates must be close. Relies on BFO for good reception; otherwise it is receiver quieting. If Earhart were using wrong antenna, then swinging the loop would have no effect.

Fundamental limit: λ = 300Mmps / 7.5 MHz = 40 meters. Give 4 meter for 0.1 λ. 10 inch loop is 1 meter, so 4 turns. Calculate inductance and resonate with 200 pF. Large loops on ships would be a problem. Where was the picture of those loops?

Scenarios[edit]

Lead in issue. Practice at the time (from RA-1B manual) was single wire lead in made short as possible. That would fit with Gillespie having transmitter in cabin near Vee antenna connection. Too much trouble to have ventral leadin go to roof to reach coupler and then come back to floor. Consequently, R is dysfunctional, T D B essentially identical. No unidirectional determination.

One modified WECo receiver. Is there an A/DF switch? How was TA connected. There was a LF antenna (bands 1 and 2) and HF antenna (bands 3 and 4). Band 1 188 to 420, Band 2 485 to 1200, Band 3 1500 to 4000, Band 4 4000 to 10000 (is this where Long believes RA-1 stops at 10 MHz?). https://tighar.org/Projects/Earhart/Archives/Research/ResearchPapers/ElectraRadios/Figure1.html So TA could couple into LF and ventral antenna to HF (or dorsal antenna with breakin). Must route HF through coupler because there is no DF setting.

One prototype Bendix RA-1 receiver. FA/TA/DF switch. Probably not remote controlled, so would have to go aft to flip the switch. Gurr enamored of receiver, but could he be enamored of WECo as well? Was Gurr asked in the 1970s? What if AE trained on RA-1 (use DF switch), but then RA-1 removed (no DF switch on WECo)?

Both receivers. Only search for null because receiving antenna would be connected to WECo.

Directionality of antennas. Fly around airport at Lae would be both close and broadside. Receiving antenna pointed at target would at null? Broadside would be collecting atmospherics.

Would turning plane onto sun line give directional increase?

Earhart did try receiving 500 kHz at Lae.

Only time Earhart determined a minimum was at Burbank under direction (but also flying to Hawaii). Gurr expected her to improve her skills during travels. But if in March, she was heading for trouble. Did RDF work going into Honolulu? (Advantage of high peaks; probably familiar to Gurr and Manning; Manning could operate equipment.)

Explanations for no minimum. If coupler has antenna connected and T not peaked, then omni could dominate and there would be no minimum.

Training for second flight. Yes, a lot more legs, but RDF failed going into Africa and may not have been repaired until Darwin. Likely that broken before Africa and not repaired until Darwin. If repaired before Darwin, then it broke again. Alternatively, AE failed to use equipment properly closing Africa; equipment broke after Africa but before Darwin; equipment likely to break again. AE watched while tech took bearing on ground; no indication of training AE. Darwin repair comments have been used to support 2nd receiver, but it may have just been the coupler. The failed generator is a twist. No fuse in couplers. Dynamotors may have had fuses. But coupler would have run off receiver dynamotor.

Long says remote for RA-1 at copilot station, but image shows in pilot station. (Miami photo).

Coupler designs. Reactance: 14154.281670205

The simple view. WECo transmitter works and is connected to dorsal Vee. Everybody hears 3105. 6210 reportedly has problems at Lae and nobody hears it after flight. Maybe Vee has problems at 6210. Only the WECo receiver. RDF connected to LF terminal and only used for bidirectional bearing. That means anything below 1400 is heard only through the loop, and that would be the right thing for RDF to radionavigation beacons and broadcast stations. It does require that the BFO be switched on for radionavigation beacons. Radio repair technicians at Lae probably took bearing on 500 kHz signal at Sal. while Earhart watched. Earhart failed to take bearing on 7500 at Lae because loop not connected to receiver in bands 3 and 4. She could turn loop all she wanted and not detect any change.

Earhart should have been able to take bearing on 500 kHz beacon transmitted by Itasca. With simple WECo connection, selecting tuning band would get correct antenna. But operator still needs to turn on RDF, select proper band and function (anything but R) on RDF coupler, turn on BFO, and turn off AVC. If fuse had blown again, then the loop would not work at all.

If ventral antenna wiped off at Lae, that would explain no HF reception. But it causes a huge problem explaining the 7500 kHz report. If there were two receivers, then the loop would always be connected to it, and they may have only tried to receive 7500 on that receiver (never looked for 3105 or 6210 or 500. Attempt would have been compounded by not using BFO. For 7500 kHz test at Lae, were they using voice? Failure to obtain a minimum implies receiving a signal; Earhart thought the signal was too strong. Unlikely for Lae to have 7500 kHz transmitter; did it have voice on nearby frequency?

If radio comm worked when RDF fuse was blown, then omni antenna not connected to RDF (one receiver model).

Earhart confused about equipment. Earhart thought whistling was necessary. If she did not do Morse code, then BFO would be foreign.

(Hey, switching to CW would disconnect Vee!)

Long and Long have nice section showing the Captain of the Itasca would have few worries because there were multiple ways for locating the Itasca.

  1. Itasca would be transmitting the requested 7.5 MHz homing signal Earhart requested. (Check this.) Earhart wanted voice transmissions, but Itasca could only do Morse code at that frequency. Still, it would be easy to RDF Morse code even if one could not copy it. Once Earhart was close, she could get an RDF bearing and head straight for Itasca. Earhart heard the 7.5 MHz transmission, but she could not find a minimum that would determine the bearing. She had similarly failed to find a minimum during a test at Darwin. It is likely that her RDF equipment was not designed to work at 7.5 MHz. (She could have been misled by the band numbering.)
  2. Itasca would be transmitting a MF beacon (375 kHz or 500 kHz). Earhart did not request this, but Manning had requested these transmissions before the crash at Luke Field. It is not clear if Earhart ever monitored these frequencies. She was not skilled at radio and may not have recognized that removing the trailing wire antenna did not destroy the RDF loop's ability to receive the MF beacon. Several authors suggest that even if she had tried, the numbered band mismatch between the RDF coupler and the RA-1 would have thwarted the attempt.
  3. Itasca believed Earhart could transmit on 500 kHz, a frequency that was within the range of Itasca's RDF equipment. Itasca could take the bearing, tell Earhart, and she could fly the reciprocal. Earhart had discarded the 500 kHz trailing wire antenna after the Luke Field crash, so she did not have an efficient antenna. Gurr had lengthened the vee antenna to give some 500 kHz ability, but it would have been very poor. Gurr's modification may have been removed in Miami. Neither Earhart nor Noonan were authorized to transmit on 500 kHz. It is unlikely that they attempted a 500 kHz transmission; Itasca did not hear one. (Check.) Even if Itasca could have obtained such a bearing, it could not have told Earhart that bearing because they were not having reliable communications.
  4. There was an experimental high-frequency direction finder set up on Howland Island. It would be able to get a fix on Earharts 3105 and 6210 kHz transmissions. Unfortunately, the HF direction finder was powered by batteries, and it was turned on while Earhart was far away. By the time a fix was important, the batteries were depleted and the direction finder was useless. Furthermore, even if a fix was obtained, Earhart was not copying Itasca, so she could not be told which direction to head.
  5. Itasca was going to generate smoke. Howland Island is essentially flat; it is not visible over the horizon. Generating smoke should give a high altitude target that would be visible over the horizon. (During World War II, submarines would detect freighters over the horizon by their smoke.) On the day, the weather was clear, but Itasca could not generate a decent smoke column. Instead, the smoke did not rise very high, and it dissipated quickly. In addition, Earhart's plane ended up where it was cloudy; instead of flying at a high altitude with an extended horizon, she flew under the clouds at 1000 feet. The horizon would be close. Earhart was not close to the Itasca; it was clear around the Itasca but cloudy around Earhart.

Bands[edit]

Band Radios Radio Direction Finders
20B
(modified)		 RA-1(A) RA-1(B) RA-1B RA-1J RDF-1-A
L1130A		 RDF-1-B
???		 RDF-2-A (AE)
L1678A		 RDF-2-A
???		 DU-1
CRR-50061
1 0.188 0.420 .150 .??? .150 .??? .150 .315 .150 .315 0.500 0.590 ??? ??? .200 .??? 0.200 0.400
2 0.485 1.200 ??? ??? .315 .680 .315 .680 0.590 0.700 ??? ??? ??? 0.400 0.800
3 1.500 4.000 .??? 1.500 .??? 1.500 .680 1.500 0.680 1.500 0.700 1.050 ??? ??? ??? 0.800 1.600
4 4.000 10.000 1.800 ?.??? 2.500 ?.??? 1.8 3.7 2.500 5.000 1.050 1.570 ??? ??? ??? NA
5 NA ??? ??? 3.7 7.5 5.000 10.000 1.570 3.000 ??? ??? ?.??? 1.430 NA
6 NA ?.??? 15.000 ?.??? 20.000 7.5 15.0 10.000 20.000 3.000 5.400 ??? NA ??? NA
7 NA NA NA NA NA 5.400 8.000 ??? NA NA NA
  • The Bendix receiver was apparently a prototype, too? Bendix started in 1936.
    "The new Bendix receiver employed a high-quality, state-of-the-art eight-tube superheterodyne design, with and enhanced ability to reject unwanted signals. ... The receiver had five bands and, with the exception of 1500 to 2400, any frequency between 200 and 10,000 kilocycles could be tuned." (Long & Long 1999, p. 63)
    This statement suggests a close match with later Bendix 6-band receivers with the last band (covering 10 MHz to 20 MHz) dropped. Long and Long cite to the Bendix manual for the RA-1 series, page 2. (Long & Long 1999, p. 259) Bendix literature in 1941 gave different limits. Is there just an RA-1 receiver (no A or B suffix)?
  • Manning gave contact instructions for the Hawaii to Howland flight during the first attempt; it had < 500 kHz RDF reception. Manning sent the following message to US Coast Guard Lieutenant Johnson:[5]

QUOTE FOR CUTTER AT HOWLAND ISLAND SIX HOURS AFTER DEPARTURE OF PLANE FROM HONOLULU CUTTER PLEASE CALL THE PLANE FROM 1 TO 6 MINUTES PAST THE HOUR AND THE FIRST 2 MINUTES ON 3105 KCS THE 2ND 2 MINUTES ON 6210 AND THE 3RD 2 MINUTES ON 500 KCS UNTIL CONTACT IS ESTABLISHED PERIOD IF NO CONTACT MADE AFTER 3 HOURS CUTTER SHOULD TRANSMIT MO LONG DASHES ON 375 KCS FOLLOWED BY CALL LETTERS FOR PLANE TO TAKE BEARINGS PERIOD THIS TRANSMISSION SHOULD BE SENT EVERY TEN MINUTES COMMENCING ON THE EVEN HOUR LASTING 4 MINUTES ATTEMPTS WILL BE MADE BY THE PLANE TO CONTACT THE NEAR SHIP AFTER THE MO TRANSMISSION IS FINISHED IN THE MANNER DESCRIBED ABOVE SHIP SHOW SEARCHLGHT AS PLANE APPROACHES ISLAND DURING DARKNESS AND MAKE SMOKE DURING DAYLIGHT UNQUOTE

  • The transmissions on 375 kHz (KCS) were for RDF, and they imply that Manning believed his RDF equipment could receive those transmissions. The RDF-1-A had a lower limit of 500 kHz.
  • WECo 20B from https://tighar.org/Projects/Earhart/Archives/Research/ResearchPapers/ElectraRadios/ElectraRadios.htm
  • RA-1(A) and RA-1(B) from http://aafradio.org/docs/Bendix_1941_brochure.pdf (gives consistent range -- 6 band -- does not stop at 10.0 MHz!)
  • RA-1B and RA-1J from http://www.tuberadio.com/robinson/Manuals/RA1B.pdf Copyright 1941, dated January 1942
    Instruction Book for Models RA-1B, RA-1I and RA-1J Aircraft Radio Receiving Equipment (PDF). Baltimore, Maryland: Bendix Radio. January 1942.
    page 2 comments that antenna connection should be a single wire, 16 to 18 BS gauge, spaced away from the fuselage. The goal was to minimize capacitance; a shielded conduit was not recommended. "The ideal installation would have the receiver connected to the antenna lead-in by means of a single insulated conductor suspended in air throughout its length." The connection is a binding post.
    page 3 grounds the receiver to the fuselage
    Binding posts: A (for fixed or trailing antenna), DF (direction finder) and G (ground).
  • DU-1 from http://aafradio.org/docs/DU-1.pdf (later model used in WW II; uses transformer to multiply variable capacitor)
    Type CRR-50061
    http://www.bendixradiofoundation.com/comp_program.htm shows projects
    DU-1 1940-1942
    Project database
    Shows that project number 135, "DU, DW, DV Direction Finder", started 8 February 1938, Section Chief was Brais, and closed 1 November 1939
    See also project number 1044, "Preparation of Bid - Navy Direction Finders", started 2 November 1939, Section Chief Eltgroth, and closed 22 November 1939
    "Engineering Projects 1937-1953 - Bendix Radio Foundation", https:www.bendixradiofoundation.com/documents/BxRadEngProj.xls retrieved 19 June 2018 (Excel spreadsheet)
    DU-1 Manual shows shielded cable connection
  • RDF-1-A from http://aafradio.org/docs/1936_NRL_Aircraft_Radio_Training_Manual_DF_Chapter.pdf
    Type L1130-B
  • RDF-1-B from Long & Long (1999, p. 63) (see also notes) (is Long guessing from RDF-2-A manual?)
  • RDF-2-A manual, 1 February 1937, cited in Long & Long (1999, p. 267)
    Believe Elgen Long had access to Bendix documents. The document is in a listing:
    Bendix Radio Foundation Print Media Data Base --Revised-4/01/2017
    https://www.bendixradiofoundation.com/documents/Print%20Media%20170401.xlsx (Excel)
    line 1381: 1379, Box 3.02.0.09, Folder 7.01, , RFT-2 Aircraft Direction Finder
    line 1382: 1380, Box 3.02.0.09, Folder 8.01, , RDF-2-A Aircraft Direction Finder
    Bendix Radio Foundation
    http://www.bendixradiofoundation.com/coll_drawing.htm
    http://www.bendixradiofoundation.com/documents/Drawings%20170317.xlsx
    line 313: RDF-2 DIRECTION FINDER, 12-10-1936, "RDF-2,DIRECTION,FINDER,L1678A,AMELIA EARHART"
    line 314: RDF-2 DIRECTION FINDER, 12-10-1936, "RDF-2,DIRECTION,FINDER,L1678A,AMELIA EARHART"
    Microfiche database
    1938 Direction finding report MZ-100A
    1939 Direction finding report MZ-100A
    1940-42 DU-1
    June 1937 Direction Finder CSX
    several MN-* series
    http://www.bendixradiofoundation.com/documents/Malcom%20Taylor's%20at%20Bendix.pdf;
    "There he started work designing the coils for the superheterodyne six-band receiver prototype which was used by Amelia Earhart during her fatal 1937 flight . While on this project he was assisted by Karl Finsder, a very knowledgeable radio technician. By 1938 this radio became the RA 1 with models RA 1B, RA 1I and RA 1J with remote control dynamometer and was perfected through shock and exposure testing. The six bands covered the range from 0.15 to 15.0 MHz. They built about 100 units for US Navy aircraft installation and also some units for the Argentine Navy. The RA 1 was later farmed out to the Sparks Worthington Company of Jackson, Michigan, a former producer of home radio receivers. Malcolm donated a copy of the RA 1 Technical Handbook to the Bendix Radio Foundation for the museum archives at 1415 Key Highway in Baltimore, MD."
  • notes for page 63 cite limited space for band info, RDF-2-A page 12, cited in Long & Long (1999, p. 259)
  • listened to WQAM 560 kHz in Miami but heard nothing Long & Long (1999, p. 135)
  • listened to KGU (AM) 760 kHz in Hawaii but that was set up by Manning
  • Elgin Long:
    The loop coupler control box had been custom-made to electronically adapt the standard Navy RDF-1-B direction-finder loop to the new Bendix RA-1 receiver. Five frequency-band numbers (labeled 1-5) were shown on the face of Earhart's loop coupler, but the Navy model had six bands...the radio direction-finder loop was designed to take bearings only on frequencies between 200 and 1430 kilocycles, whereas the receiver could receive...up to 10,000 kilocycles...If Earhart or Manning tried to take a bearing on a high-frequency signal using the Bendix receiver's upper bands, the resonance of the loop would be so far out of tune they probably could not get a minimum signal."
  • GP believes RDF went down to 200 kHz:
    June 25th at 2245GMT, the Coast Guard San Francisco office sent a radio message to Public Domain This article incorporates public domain material from websites or documents of the United States Coast Guard.[6][7][8]
    ... Mr. Putnam now at Oakland and advises Ms. Earhart at Bandoeng Java for repairs to motors and departure indefinite. She will cable details communications from Port Darwin direct San Francisco and you will be given all information immediately. All communications from plane will be on 500, 3105, or 6210 kHz by voice, positions being given at 15 and 45 minutes past the hour. Itasca adjust transmitter for possible use 3105 kHz for voice. Direction finder on plane covers range of about 200 to 1400 kHz.
    Public Domain This article incorporates public domain material from websites or documents of the United States Coast Guard. logs also have USCG SF to Itasca 26 June 1937:[8]
    FOLLOWING INFORMATION FROM EARHART THIS DATE -
    "HOMING DEVICE COVERS FROM 200 TO 1500 AND 2400 TO 48 KILOCYCLES. ANY FREQUENCIES NOT, REPEAT NOT, NEAR ENDS OF BANDS SUITABLE." SUGGEST USING SUITABLE FREQUENCIES HAVING IN MIND UNCERTAIN CHARACTERISTICS OF HIGH FREQUENCIES. USE 333 KILOCYCLES OR FREQUENCY IN THAT VICINITY AND TRY 545 KILOCYCLES AFTER TESTS WITH STATIONS YOUR LOCALITY TO DETERMINE WHICH IS BEST. ADVISE IF IMPOSSIBLE TO PLACE TARE 10 TRANSMITTER ON 3105 KILOCYCLES. EARHART AT LAE VIA TUTUILA EXACT FREQUENCIES SELECTED AND ASSUME CONTINUOUS SIGNALS AFTER HER DIRECTION FINDER IN RANGE. SEE BROADCAST ON QUARTER AFTER AND QUARTER BEFORE HOUR ON 6210 AND 3105 KILOCYCLES.
    AM ADVISING EARHART THAT ITASCA WILL VOICE RADIO HER ON 3105 ON HOUR AND HALF HOUR AS SHE APPROACHES HOWLAND. REPAIRS MADE AND EARHART NOW AT SOURABAYA. EXPECTS LEAVE DAWN THIS DATE FOR PORT DARWIN AND NEXT DAY FOR LAE. ADVISE PRIORITY IF ADJUSTMENTS TARE TEN TRANSMITTER SATISFACTORY FOR USE ON 3105.
    Speculate: That gives a quasi-consistent statement (if we take "48 kilocycles" to be "4800 kilocycles", but it suggests that it was driven by the receiver's bands rather that the RDF bands. There are other statements that 200 KHz was receiver's low end, and the receiver has a gap around 1500 to 2400.
    Speculate: maybe it does use the transformer, but the capacitor didn't do a factor of four, so it used 5 bands to try to cover the DU-1's later 200 to 1600 kHz.

All of this leaves me confused. On 25 June, Putnam states RDF is 200 to 1400 kHz. On 26 June, Earhart specifies RDF 200 to 1500 and 2400 to 4800. Earhart is not saying 7500. Sounds like she is saying 333 or 545 kHz. What evidence for 7500? Could Earhart state the range in one place but not apply the range at another place? It should at least raise red flags on Itasca.

The 7500 was a last minute radio request at 0758 (Itasca); she responded at 0803, but wanted results set on 3105 kHz.

Were transmissions too short for adequate bearing? Request for long count is to aim DF antenna.

Earhart's test at Darwin was probably at Darwin's ordinary transmitter frequency and not 7500. When was she in Darwin? That would be after Java. Did she evaluate the Darwin test and realize LF or homing 3105 would work? Is the failure that 3105 kHz was not suitable? If she expected Itasca to transmit on 3015 and 6210, then why not home on those signals and forget about 7500? If two receivers, then they could be independent, but with one receiver, the retuning does not make sense.

http://earchives.lib.purdue.edu/cdm/compoundobject/collection/earhart/id/3119/rec/55 page021 page023. AE requested 7.5 MHz! Report notes contradiction. Batteries ran down, but not whole story; Earhart's transmissions were voice (not constant amplitude) and brief (never more than a 8 seconds). page044. Smoke remained concentrated and did not thin out greatly! Where was that picture? Interesting conclusions on page048: Earhart had receiver troubles; ack for 7500 may have been other signals; possible sun line but no position.

Gas tanks[edit]

Another screwy issue is the gas tanks providing floatation for the plane.

Paul Mantz, technical adviser to the aviatrix, who twice crossed the Atlantic, said in Burbank, Calif., the plane's six gas tanks would give it buoyancy to stay afloat "indefinitely."[9]

Lockheed had a different conclusion: the plane would sink quickly.

Gross weight is 16,000 pounds (Johnson). Fuel tanks 1150. 9545 Oil tank is 50? 415.

Long and Long also suggest the fuel dump valves would let water in.

Sources[edit]

Flight to Hawaii[edit]

Flight to Africa[edit]

Intended to reach Dakar, Senegal, but they ended up north of their desired position and went to Saint-Louis, Senegal instead. Noonan didn't know why. He passed a note.

AE's book[edit]

Oakland to Honolulu on first attempt:

"Keep the Makapuu beacon ten degrees on the starboard bow," he [Fred Noonan] ordered.
What he meant was that I should tune my Bendix radio direction finder to indicate the location of the beacon, and then head the plane as he directed. this was the first time I had used this recently developed Bendix instrument. On this Pacific hop it was one of the most interesting and valuable on board, performing perfectly.[10]

Luke Field to Howland on first attempt:

We had 900 gallons on board. That was as much as we carried comping from Oakland, although the contemplated distance to Howland was 600 miles shorter than the first leg of the journey.[11]

Confusion about Quoting C. B. Allen, New York Herald Tribune.

p. 97 left Miami shortly after 6 AM. Tuned WQAM.

p98 6:30 AM Great Reef of the Bahama Banks 7:00 AM Andros Island

p 129 Natal to Saint Louis

About midway we passed an Air France mail plane. Unfortunately I could not "talk" to it. The mail plane's radio equipment, I believe, is telegraphic code, while mine, at the moment, was exclusively voice telephone.

p 130

Have tried get something on radio. No go. Rain, static.

p 217

@ Darwin, by the way, we left the parachutes we had carried that far, to be shipped home. A parachute would not help over the Pacific.

p 223

The monoplane is weighted with gasoline and oil to capacity.

Navigation[edit]

Bubble octant[edit]

There's a huge topic here.

Development. Byrd. P. V. H. Weems. Aircraft navigation in general. Noonan a student of Weems.

There's the octant versus sextant issue. Main source states octant, but there are several pictures that show instruments with a sextant label. Anderson provides a schematic of a bubble sextant, but the scale implies octant.

The accuracy issue and the requirement for averaging.

Then there is what happened to the octant that Manning borrowed from the Navy. I think Long and Long had him return it before the second attempt. Long p. 72: Noonan wanted an octant but had not obtained one. Commander Ragsdale asked North Island Naval Air Station in San Diego if one could be borrowed. Long p. 73: The Secretary of the Navy approved the loan; Manning would have to sign for it. Long p. 103: Manning, who was personally responsible for the loaned bubble octant, retrieved it from the rear of the Luke Field plane crash.

Warner gives a detailed history of the instrument. Many companies are involved: Brandis, Bausch & Lomb, Pioneer, Keuffel & Esser, Bendix. Victor Carbonara, an Italian designer hired by Brandis, designed the Brandis air octant Model 206 in 1925. Warner page 97: "Carbonara realised that air navigators never measured angles greater than 90°, and so could use an octant rather than a sextant." Warner page 104 speculates that Noonan used a Pioneer sextant.

Mapframe[edit]

Map
Earhart Lae to Howland. Cut version; modular longitude. Preview shows antimeridian, Howland, and route from the antimeridian to Howland. Clicking through paints details across the antimeridant 360 degrees away.


Map
Earhart Lae to Howland. Eastings version (Howland at 183). Shows antimeridian. Preview does not show Howland or the route from the antimeridian to Howland. Clicking through works.
Map
Earhart Lae to Howland. Scissor cut with MultiLineString. Same problems as first?


Try a mapframe from mw:Help:Extension:Kartographer. Also mw:Extension:Kartographer. https://tools.ietf.org/html/rfc7946 https://www.mapbox.com/maki-icons/

There are some problems with lines that cross the antimeridian. On the ordinary display, the western hemisphere lines do not appear; click on the map and they appear. I tried to game it with a longitude of 183, but it does not work. Want to intersect (159.400, -4.517) to (183.383479, 0.807179) with the antemeridian. Near the equator, the world is quasi flat; use linear interpolation to find the latitude:

But that does not do what I want either. The ordinary page display works correctly now (both hemispheres display), but click on the map and the western hemisphere lines disappear. Try using -179.999; does not help. Also, Howland Island disappears. Coordinates are messed up. Using continuous coordinates (Riemann) confounds ordinary display but clicking gives good map. Just shoot me.

GeoJSON suggests cutting geometry at the antemeridian instead of crossing it. See RFC 7946 §3.1.9.

Isolated test[edit]

Tracked in Phabricator
Task T313615

Setting longitude to -179 or 179.

Map
try several options

Mumble[edit]

The flight's route is not known. Although Earhart radioed her position from time to time, those reports are not clear.

Although taking a great circle route would minimize the distance traveled and presumably save fuel, there are other considerations.

Long & Long believe they deviated south to avoid some heavy weather. Long supports his theory with a high altitude claim. Long points to a typical 10 percent error in travel.

One radio message copied the position as 157.

The position is too close to Lae to make sense. Long suggests it is the position as of the last fix.

Gary LaPook suggests the position is 150 7'.

LaPook is an attorney specializing in aircraft accidents; he is a celestial navigation instructor and former commercial airline pilot.[12] LaPook believes the crash and sink theory, and did not believe the aluminum found on Nikumaroro belonged to Earhart's Electra. He believes that if Earhart were south of Howland, the weather was clear, Noonan could have taken a fix, and they could have headed for the closer Howland rather than far away Gardner.

Van Asten takes that much further. That Noonan, instead of taking a great circle, went island hopping. Although the Pacific is vast, there are islands and reefs. During daylight, if one sees an island, then one knows his position without doing any difficult celestial navigation. Instead of taking the shortest route, a great circle route, (and fighting the weather), Earhart flew toward landmarks that could be used for navigation. Van Asten believes the Electra was at Nukumanu Islands with a firm navigation fix around dusk. At night, one can take some star fixes to determine position, but one can also look for lights on islands.

On June 30, Earhart received from Nauru:[13]

New Nauru fixed light 0.32°S, 16.55°E [sic 166.93 or 166 55'], 5000 candlepower, 5600 [sic – 560] above sea level, visible from ships to naked eye at 34 miles. Also, there will be a bright lighting all night on island from phosphate field workings. Weather 8AM: Baro 29.908, therm 84, wind SE at 3, fine but cloudy, sea smooth to moderate.

Need to check the message. The gross longitude error is easy to recognize. The fine error might be convention, but should also be easy to resolve: Nauru is a few miles wide; an error of 0.5 degree is 30 miles. The 5600 feet was an error; the height is only 560 feet (170 m).[14] The beacon height would give Nauru steep sides: a few miles wide and a mile tall. That could have some impact on visibility, but at 10,000 feet, the light would be visible from XXX miles. Clouds could obscure it. The 34 miles is due to the beacon's height and the height of a typical bridge. At 132 (4 times as far), the light intensity would be 1/16th. Binoculars would make up for inverse square loss, but not for scattering loss. Van Asten believes Noonan would have headed for Nauru; if they saw the loom of the industrial lights or the beacon, then they could use it for a line of position. Nauru is enough out of the way that they would not fly all the way to the island. At some point before reaching Nauru, they would turn west to head toward Howland. If they saw the island's light, it would be a bonus.

Van Asten believes Noonan would have chosen a latitude line course. It would be easy to stay on a latitude line by making simple star sights. Van Asten speculates that Noonan would have chosen a latitude line that take them over either Nikunau or Beru Island. Question: looking for lights or visible at morning twilight? Twelve degrees is almost an hour, which is almost 1000 miles. I think the islands are only 500 miles away (three hours of flying). But sunrise and 100 miles from Howland does not fit; that puts the islands two hours back.

Dawn twilight

There are three versions of twilight: civil, nautical, and astronomical.[15] They occur when the center of the sun is less than 6, 12, or 18 degrees below the horizon.

The calculation looks close. Civil twilight is probably only 20 minutes (5 degrees at 15 degrees per hour). Nautical twilight (when most stars can be seen) would be another 25 minutes. When did O'Kane submerge?

LaPook points out a star fix was unlikely. Earhart reported overcast at 2:45 AM, so a star fix was out.[12]

Van Asten claims the project times match up well (but he can fiddle with the air speed to make them come out right).

Van Asten claims one fix was reported before it could have been made. Need to understand that argument. He offers some explanation about navigators doing their homework (precomputing some values) so they don't have to do the work during the flight.

Locations from reports. Here's a quotation:[16]

Amelia’s first intelligible message wasn’t received at Lae until more than four hours after her departure. At 2:18 p.m., Lae radio operator Harry Balfour heard, “HEIGHT 7000 FEET SPEED 140 KNOTS and a remark that sounded like “EVERYTHING OKAY.’” Balfour sent weather reports until 5:20 p.m., but none were acknowledged by Amelia. At 3:19 p.m., she reported, “HEIGHT 10000 FEET POSITION 150.7 east 7.3 south CUMULUS CLOUDS EVERYTHING OKAY.” At 5:18 p.m. (0718 GMT), her position was “4.33 SOUTH 159.7 EAST HEIGHT 8000 FEET OVER CUMULUS CLOUDS WIND 23 KNOTS.” This put the Electra just southwest of the Nukumanu Islands, on track about one-third of the way on course to Howland.

Table of locations and projected times.

Radio schedule with Lae would be 18 minutes past the hour. Find reference. AE not heard for 4 hours. Why? AE would be on daytime frequency. When did she announce switch? Lae was upset because she came in loud and clear.

Radio schedule with Itasca was... Itasca picked her up during the night on... Near Howland, AE switched freqs and never heard on new freq. (Long surmises instant of gasoline problems.)

Location Coordinates Time Comments
Lae Airfield 06°43′59″S 146°59′45″E / 6.73306°S 146.99583°E / -6.73306; 146.99583 0000Z origin
scheduled radio 0018Z 10:18AM Lae not heard
scheduled radio 0118Z 11:18AM Lae not heard
scheduled radio 0218Z 12:18PM Lae not heard
scheduled radio 0318Z 1:18PM Lae not heard
Harehare, Shortland Island 7°08′22″S 155°47′39″E / 7.1393391°S 155.7942539°E / -7.1393391; 155.7942539 0418Z 2:18PM Lae Van Asten guess
Earhart radio msg 7°18′S 150°42′E / 7.3°S 150.7°E / -7.3; 150.7 0519Z 3:19PM Lae Balfour
Choiseul Island 7°3′S 157°0′E / 7.050°S 157.000°E / -7.050; 157.000 LaPook, Van Asten
Earhart radio msg 4°20′S 159°42′E / 4.33°S 159.7°E / -4.33; 159.7 0718Z 5:18PM Lae Balfour
Nukumanu Islands 4°31′S 159°24′E / 4.517°S 159.400°E / -4.517; 159.400
Nauru radio 1030Z Nauru radio received "Ship in sight ahead" at 1030Z.[17]
USS Ontario 2°59′S 165°20′E / 2.983°S 165.333°E / -2.983; 165.333 1030Z TIGHAR 100 knot GC route[18]
MV Myrtlebank 2°20′S 167°10′E / 2.333°S 167.167°E / -2.333; 167.167 1030Z TIGHAR 130 knot GC route
Nauru 0°32′S 166°56′E / 0.533°S 166.933°E / -0.533; 166.933 Van Asten
long run
Tabiteuea 1°21′S 174°48′E / 1.350°S 174.800°E / -1.350; 174.800 Van Asten
Beru Island 1°19′S 175°58′E / 1.317°S 175.967°E / -1.317; 175.967 Van Asten
Nikunau 1°21′S 176°27′E / 1.350°S 176.450°E / -1.350; 176.450 Van Asten
head to Howland
Cross Dateline July 3 back to July 2
Earhart radio "200 miles out" 6:14AM Itasca radio log
Earhart radio "100 miles out" 6:45AM Itasca radio log
Howland Island 0°48′26″N 176°36′59″W / 0.807179°N 176.616521°W / 0.807179; -176.616521 destination

Lae. Harehare. Nikumanu.

Although a great circle route or a rhumb line route is the direct approach, several authors have suggested other routes. Long believes Earhart flew south of the great circle route to avoid some weather. Van Asten believes, "Noonan preferred to fly via as many as possible identifiable landmarks, islands and other contingent aids to navigation, to avoid the hazard of going astray from a long (2,556 mls, 4,113km) great circle route, partially to be flown in adverse weather."[19]

Van Asten believes the route was Lae to Ghaghara, Choisel Island.[20] From there to Nikumanu Islands (where there would still be light to see the islands). From there, the flight would head toward Nauru Island. Nauru has a lighthouse, and van Asten believes the light would be visible from far away. "Nauru (visual range 135 mis from 12,000 ft) when Amelia announced to have the glare of the island's industrial lights and the 560 feet high fixed light in sight."[21] Van Asten believes Earhart would have turned due East before reaching Nauru. By traveling on the 1 degree 23'S meridian, navigation would be simplified. Star fixes could be used to stay on the meridian, and the route would pass over Tabiteuea, Beru, and Nikunau in the morning. From Nikunau, Earhart would head to Howland Island.

Where is Howland Island? Actual. AE thought:[22]

By consulting Lippincott's Atlas of the World at the Hydrographic Office in the beginning of 1937, navigational consultant C. Williams copied the Howland coordinates 176°43'-W; 00°49'-N which is a position 5.8 mls westwards of the true coordinates pair (176°38'-W;00°48'-N).

Do they fit in with sighting the ship?

Then Earhart would know her position and be able to head for Howland, a hop of less than 400 miles. Flying with dead reckoning, the error near Howland would be 40 miles.

At sunrise, they would know they were on a sun line, but it would not tell them much about north or south. It also seems that no observation need be done beyond recognizing the break of dawn and noting the time. One could measure the azimuth, but I think a small error there translates to huge distances.

After sunrise, the sun can still be used for further sun lines. The easy method is just advance the sun line by the distance you think the plane has traveled. At 200 miles out, the error might be 20 miles.

Gary LaPook and H. A. C. van Asten disagree. LaPook says don't bother reading Van Asten's EJN articles. They get into a navigation debate. LaPook fails to understand the difference between measuring the upper limb of the sun (first light) and the central position of the sun. The sun subtends about 1 degree, so the error is significant. They also debate the use of instruments. Apparently the sextant always matches the lower limb because that method is the best for accuracy. One just adjusts the sextant to put the lower limb at the horizon. The bubble octant is apparently different. One probably centers a star on the bubble or the bubble on the sun or moon. Consequently, Van Asten seems to have the better arguments. Van Asten did make an error, but that error does not destroy the rest of his argument. There's a debate about the equation of time that needs more investigation. Van Asten seems to wave off altitude considerations. Is taking a shot at 1000 feet the same as at sea level? Van Asten implies altitude is an issue, but thinks it does not affect the calculation significantly?

Plug in numbers at USNO.

Find a text about computing sun lines. Calculate the sun position for plausible locations.

Moon was not up.

This diagram of the Sun at sunrise (or sunset) shows the effects of atmospheric refraction.

At time of sunrise for Howland (1745 UTC), the sun is about 1 degree below the horizon.

NASA says the subtended angle is 0.53 degrees. https://er.jsc.nasa.gov/seh/math30.html See Sunrise; apparent sunrise when 50' below horizon, so the calculation fits. The error above is 4 arcminutes, but I did not use the most accurate coordinates. The sun moves at 360 degrees/24 hour, 15 deg/hour, 15 arcminutes/minute. The 4 arcminute error is about 15 second error. One degree is about 60 nm, so one arcminute is about 1 mile. 1000 mph is 16 mi/min or 0.25 mi / sec; 16 seconds would be a 4 mile error.

The USNO website wants position at sea level.

Itasca log (not the radio log): http://coastguard.dodlive.mil/files/2012/07/Earhart_001.jpg

This document gives 6:14 as 200 miles out, and then 6:45 position as 100 miles out -- an absurd 200 miles/hour. Long is not concerned with that discrepancy.

TIGHAR tries to explain the discrepancy away, but at 200 miles out, sunrise would not have happened yet, so position would be dead reckoning.

The next transmission could have seen sunrise within the last half hour (transmission schedule). Sunrise would be 80 miles off the line. Wait: 1 hour is 1000 miles. One minute is 16 miles. If 200 miles out, then sunrise delayed 7.5 minutes; if 100 miles out, then sunrise delayed 1.25 minutes. Sort of makes sense: sun moving at 1000 mph means 200 miles is only 1/5 of an hour. Would expect report to say 200 miles out; then, 1/2 hour later at 150 mph, to say 125 miles out. Subsequent sun fix may have improved the position estimate by 25 miles. If 100 miles out, then 1/2 hour later (7:15), should be 25 miles out. The 7:45 report would be on the line in either case (except for winds aloft issue that would leave them short). But if we are talking nautical miles, then things are slightly different.

The Coast Guard believed Earhart and Noonan went down northeast of Howland. The sky was clear at Howland, but Earhart reported clouds. That explains why she was flying at 1000 feet: to stay below the clouds. The Coast Guard believed the only clouds were to the northeast. The Itasca was making significant smoke. Earhart could not hear Itasca, so Itasca could not tell them the sky was clear around Howland.

Eleanor Roosevelt[edit]

Henry Morgenthau Jr. was a neighbor and friend of Franklin D. Roosevelt. He was Secretary of the Treasury from 1934 to 1945. At that time, United States Coast Guard was under the Treasury Department.

Paul Mantz wrote to first lady Eleanor Roosevelt on 26 April 1938 and asked for her assistance in obtaining Itasca's radio logs and its report about Earhart. Mantz knew that the first lady and Earhart had friends. Eleanor forwarded Mantz' letter to Morgenthau on 10 May 1938. Morgenthau called Eleanor Roosevelt on 13 May 1938, but she was not in, so Morgenthau talked with Eleanor's secretary, Malvina "Tommy" Scheider.[23] Morgenthau stated:[24]

Hello, Tommy. How are you? This letter that Mrs. Roosevelt wrote me about trying to get the report on Amelia Earhart. Now, I've been given a verbal report. If we're going to relase this, it's just going to smear the whole reputation of Amelia Earhart, and my ... Yes, but I mean if we give it to this one man we've got to make it public; we can't let one man see it. And if we ever release the report of the Itasca on Amelia Earhart, any reputation she's got is gone, because — and I'd like to — I'd really like to return this to you.

Now, I know what Navy did, I know what the Itasca did, and I know how Amelia Earhart absolutely disregarded all orders, and if we ever release this thing, goodbye Amelia Earhart's reputation. Now, really — because if we give the access to one, we have to give it to all. And my advice is that — and if the President ever heard that somebody questioned that the Navy hadn't made the proper search, after what those boys went through — I think they searched, as I remember it, 50,000 square miles, and every one of those planes was out, and the boys just burnt themselves out physically and every other way searching for her. And if — I mean I think he'd get terribly angry if somebody — because they just went to the limit, and so did the Coast Guard. And we have the report of all those wireless messages and everything else, what that woman — happened to her the last few minutes. I hope I've just go to never make it public, I mean — O.K. — Well, still if she wants it, I'll tell her — I mean what happened. It isn't a very nice story. Well, yes. There isn't anything additional to something like that. You think up a good one. Thank you.

In addition, there was an undated note on White House stationery sent to Mrs. Roosevelt:[25]

Mr. Morgenthau says that he can't give out any more information that was given to the papers at the time of the search of Amelia Earhart. It seems they have confidential information which would absolutely ruin the reputation of Amelia Earhart and which he will tell you personally at a time when you wish to hear it.

He suggests writing this man [Mantz] and telling him that the President is satisfied from his information , and you are too, that everything possible was done.

On 14 May, Eleanor Roosevelt sent Mantz a letter that followed that suggestion:

I have made inquiries about the search which was made for Amelia Earhart and both the President and I are satisfied that the information which we have received that everything possible was done. We are sure that a very thorough search was made.

On 5 July, Morgenthau told Eleanor Roosevelt that the government would release the Itasca radio logs to Mantz. On 21 July, Coast Guard Rear Admiral R. R. Wesche (*** check USCG Commandant) sent Mantz the logs.[26]

Search[edit]

Colorado arrived a week after and searched with its planes.

"When Captain Friedell of the Colorado rendezvoused with the Itasca at daybreak on Wednesday, July 7, and was able to confer directly with Commander Thompson, he became aware for the first time of the emotional content of Earhart's final messages. He learned then that the men listening on the Itasca believed Earhart when she said she was running low on fuel and had only a half hour left. From the strength of her radio signals, Captain Thompson believed Earhart came down within 100 miles of Howland."[27]

Lexington arrived on the morning of 13 July 1937 and launched 60 planes for the search. Stopped search 5 days later (28 July). The group had searched 262,281 square miles.[28]

Coast Guard Captain Thompson wrote its commandant Rear Admiral Russell R. Waesche on 18 October 1937, "As regards the search, I was very sanguine up to the time the Lockheed people stated that her pane could not possibly be landed without crashing and sinking."[29]

Fuel consumption[edit]

Gas on board is an issue for the Lae, New Guinea, to Howland Island leg of 2,556 miles (2,221 nmi; 4,113 km).

There are three different fuel amounts that are claimed. News reports shortly after the disappearance say 950 gallons, Earhart told local authorities about 1000 gallons, and Chater says about 1100 gallons.

The airplane capacity is 1150 gallons. A 950 gallon figure is had by filling all but the two front wing tanks (97 gallons each). A 1000 gallon figure is filling all but the two front wing tanks and including 50 gallons of 107? octane full that remained in one of the front wing tanks. The 1100 gallon figure is filling all the tanks except for one front wing tank that had 50 gallons of 107 octane fuel. A consistent figure is 1000 gallons. In addition, Earhart did not want extra weight (an extra 100 gallons weighs 600 pounds), and Putnam stated that Earhart was practicing take offs with 1000 gallons. (The map in Earhart's book gives the figure as ????.)

Long also accounts for fuel expansion.

How is the 900 gallon Hawaii figure reached? Don't fill one of the 149 gallon tanks and don't fill one of the 100 gallon tanks.

On the first world flight attempt, Earhart used 900 gallons for the Oakland to Honolulu leg of 2,400 miles (2,100 nmi; 3,900 km). The estimate was that gave a safety margin of 40%. When Earhart took off from Luke Airfield on the Honolulu to Howland leg of 1,900 miles (1,700 nmi; 3,100 km), she also had 900 gallons on board (and two other passengers).

If she didn't put any gasoline in the front wing tanks, she'd be about 200 gallons below the max, or 951. Put 50 gallons in one wing tank and don't fill the 70 gallon rear tank?

How much for Natal, Brazil, to St. Louis, Senegal 1,727 miles (1,501 nmi; 2,779 km)?

Earhart told officials at Lae that she would probably carry 1000 gallons.[30]

Longworth quotes Nesbit's articles in January and February 1989[31]

Nesbit`s article, ``What Did Happen to Amelia Earhart?`` appears in two parts in the January and February issues of a British magazine, Aeroplane Monthly. Nesbit, 67, a navigator for the Royal Air Force in World War II, has written seven books on air history in recent years.
But Nesbit argues that there is no evidence that Earhart had a camera aboard. Under any conditions, he says, she could not have reached Mili. The reason, he says, is in documents at the Public Records Office in London. Lae, where she took off, was in Australian-controlled territory. Because Australia was part of the British Commonwealth, she needed British permission for the flight.
In a letter at the Office to the British Director of Civil Aviation, Earhart applied for permission and said that, for the Lae-Howland leg, ``I shall carry probably 1,000 gallons of gasoline.``
This was less than the full load of 1,150 gallons, but even 1,000 gallons put a strain on the plane. Earhart`s husband, George Putnam, said she was practicing takeoffs with up to 1,000 gallons. Journalists at Lae, quoting Noonan, said Earhart took off with 950 gallons, according to contemporary accounts in the British Museum Newspaper Library in London.

Black states (also believes crash and sink):[32]

Miss Earhart was supposed to have made a refueling stop at Howland where a work force under Black's command had built a makeshift landing strip and had 18 drums of aviation fuel available.

18 55-gallon drums is 990 gallons. She could assume there would be fuel left in her tanks when she landed.

Chater says she had 1100 gallons, but that is based on his belief that all tanks being full except for a half-full wing tank with 100 octane fuel.

TIGHAR states (diagram also shows position of transmitter):[33]

When delivered, the airplane had a total of 13 separate fuel tanks:
  • Each wing had three tanks – 16 gallons, 81 gallons and 100 gallons.
  • In the cabin, forward of the Main Beam were two tall 118 gallon tanks.
  • Behind the Main Beam were five shorter tanks – 3 @ 149 gallons, 1 @ 70 gallons, and 1 @ 51 gallons [the 51 gallon is not shown and may be an error]

Total capacity (using 102 gallon wing tanks): 1151

Pretty clear that one front wing tank (capacity 81+16 gallons) was half full, then there would be about 1100 gallons. If the other front wing tank were also empty, the plane would have about 1000 gallons.

Leaving 100 gallons at Lae saves 600 pounds.

Tne Navy considered the fuel consumption of Earhart's plane. Economical airspeed was 130 knots (150 mph) and burning 45.8 gallons per hour and 24 hours of endurance. At 140 knots (161 mph) the plane would burn 53 gallons per hour and have 20.75 hours of endurance.[34]

Calculation checks: economical range is 3602.6 miles; fast range is 3341.5

Long & Long (1999, p. 233) interpret Kelly Johnson's fuel consumption information, determined the average airspeed was 160.5 mph, and the endurance to be a little over 20.5 hours.

Long & Long (1999, p. 232) buttresses the 1/2 hour of fuel statement. Thompson and seven other people confirmed the statement, and there were three contemporary written records: the two radio logs and some handwritten notes by United Press correspondent Howard Hanzlik.

Is there a simple way of doing this? Or is that what Elgen Long did? Effective distance is increased by headwind. 25 knots * 20 hours is another 500 nm. (nm v statute m issue.) At 140 knots, that's another 3.6 hours at 53 gph or 189.3 gallons. The flight leg of 2556 miles at 150 mph would take 17 hours and 780.4 gallons.

Kelly Johnson predicted a 40% reserve on the Oakland to Hawaii flight, but there was a tailwind.

Earhart predicted the flight would be 18 hours.[8]; Lae sent Itasca this message for Richard Black:

URGENT BLACK ITASCA -
AMELIA EARHART LEFT LAE TEN AM LOCAL TIME JULY 2ND, DUE HOWLAND ISLAND 18 HOURS TIME.

That would mean hitting the sun line about dawn? Check: Lae 10 AM is 0000GMT advance 18 is 1800GMT subtract 11.5 is 0630 Itasca. Close to sunrise at 6:15 Itasca time. The 18 hour flight time would mean 142 miles per hour (123 kn).

David K. Bowman's website.

  • van Asten, H. A. C. (July 2008). "Where to Search for the Earhart Lockheed Electra". European Journal of Navigation. 6 (2): 24–31. ISSN 1571-473X. (There are additions.)
  • van Asten, H. A. C. (April 2011). "Frederick Noonan Precomputed a Running Sunset Fix for Amelia Earhart's Flight from New Guinea to Howland, July 2, 1937". European Journal of Navigation. 9 (1): 28–32. ISSN 1571-473X.
  • van Asten, H. A. C. (December 2011). "Headwinds Aloft, Destination not sighted. Could the Earhart–Noonan Crew flying from New Guinea reach other Pacific Islands from the Vicinity of Howland and Baker?". European Journal of Navigation. 9 (3): 40–45. ISSN 1571-473X.

General coverage of theories

  • Strippel, Richard G. (November 1995). "Researching Amelia: A detailed summary for the serious researcher into the disappearance of Amelia Earhart". Air Classics. 31 (11): 12–20, 50–69?. ISSN 0002-2241. OCLC 723409375.

Upper limb issue

Gardner Island hypothesis[edit]

Bendix / Hooven guy was among the first to raise the possibility. He noticed the radiolocation of the post loss transmissions. He discarded the theory after the flyover didn't find a plane.

If one believes the GIH, then one must account for the fuel. If the plane were flown at economical speed, then it would have 24 hours of endurance. Kelly Johnson recommended flying at economical speed even when confronting a 25 (mph or knot) headwind. Without headwind, Earhart could make the flight in 18 hours and have 6 hours of fuel remaining. With a 25 knot headwind, Earhart flew another 500 nmi. That could add 4 hours to the trip time, but it only took another 2 hours. Therefore, Earhart flew 10 percent faster than economical speed. Also the detour to avoid weather. But economical speed also varies. Look at Johnson's report. Earhart radio messages imply Johnson's recommendations were not followed. High altitude early. High speed early. Density altitude; speed reports. In particular, one must counter Earhart's statement about 1/2 hour of fuel remaining. TIGHAR suggests the 1/2 hour of fuel remark was until they had to start using the reserve. TIGHAR also points out the plane was still transmitting an hour later; other point out a shift to maximum endurance would increase time aloft. TIGHAR also argues that Earhart did not waste time searching for Howland but immediately went for Baker and then the Phoenix group. Witnesses comment about Earhart's stress being consistent with running out of fuel.

If one believes the transmission on XXX came from Earhart on Gardner Island, then one must explain the plane staying on the island for 6 days and then disappearing in the time from last transmission to flyover. Bendix / Hooven guy discarded the theory for that reason; he later adopted the Japanese took the plane.

If one adopts the Japanese took the plane, then one must explain why. The Japanese could be heroes for finding her. If the plane was in the Phoenix Islands, then it would be an unlikely spying mission. If Earhart had film, then the Japanese could confiscate the film and still be heroes. Earhart's path went over.... Also, daytime photography would be observed; nighttime photography may not show much. Why would the Japanese be around Gardner a week after the crash while the US was conducting an intensive search?

If one believes the (non-native) skeleton is Earhart, then one must account for the failure to remark about the presence of surgical holes in the skull. Earhart had two holes made to relieve sinus issues.

If one believes the Lockheed made it Gardner, then one must explain why Howland was not seen. AE was flying at 1000 feet presumably due to cloud cover, but the sky was clear at Howland. If AE had come south from the known clouds, the sky would have cleared, she would have climbed to get a better horizon, and then Howland should have been seen. Consider Itasca making smoke and image shown in one AE book.

Was the Navy was right? Navigation put her too far north, so she was just circling there? If she had headed to the Phoenix Islands, she would have seen Howland (smoke) or Baker (low to the sea).

Any event seems to say a significant navigation error. Too far north and stuck in the clouds, or too far west so cannot see island even in clear weather.

Other people are possible castaways. The shipwreck lost 11 crew, but only four bodies were recovered.

If one believes the shoe is hers, then one must account for the shoe being size 9 while Earhart wore size 6.

If one believes the aluminum piece is from the Electra, one must account for the non-1937 aluminum grade and that the hole pattern fits a WW II plane.

Wasn't there a WW II plane crash? TIGHAR had said that a plane was visible on the island.

Fuel and flying time is an issue for the hypothesis. Earhart left Lae at midnight GMT. The "WE MUST BE ON YOU" transmission was at 7:42 am Itasca time, which would be 20:12 GMT — a little over 20 hours into the flight. If the plane followed Kelly Johnson's schedule with 1100 gallons of gasoline, then the Electra would still have four hours of fuel on board. That would be enough to make it to Gardner Island, which was a little over 2 hours away at 150 knots. However, Earhart's message also stated the plane was running low on gas and that there was only half an hour left. One explanation is that the Electra had only a half hour before it would start using its reserve supply. Certainly the plane was still in the sky an hour later when the last known transmission at 8:43 am was heard.

Speed over ground would be 127.8 miles per hour (111.1 kn; 205.7 km/h). Expected flying time at 150 mph would be 17 hours. Expected flying time at 150 kn would be 14.8 hours. Suggests headwind of 40 kn. Knew it was 27 knots at one point. One alternative is to step on the gas.

Did they want to be over Howland before dawn at 6:15? Leaving at midnight, flying for 16 hours, would put them precisely at dawn. Half an hour before dawn would be 75 nmi away and right at the horizon. They believed they were 200 nmi out at dawn.

Flight from Oakland to Hawaii was 16 hours. Speed 150 miles per hour (130 kn; 240 km/h) (distance is 2400 miles).

References[edit]

  1. ^ Long & Long 1999, p. 60 Long and Long cite to the daily journal of O. Vernon Moore; entry on 15 February 1937.
  2. ^ Long & Long 1999, p. 61
  3. ^ Long & Long 1999, p. 63 No supporting evidence cited at Long & Long (1999, p. 259).
  4. ^ Morrisey, Muriel Earhart; Osborne, Carol L. (1987). Amelia My Courageous Sister. ISBN 978-0940997004.
  5. ^ Long & Long 1999, p. 70
  6. ^ https://tighar.org/wiki/Failure_to_communicate
  7. ^ Jacobsen, Randall S. (n.d.). "The World Flight, Second Attempt: Communications & Coordination". TIGHAR. Retrieved 23 May 2018.
  8. ^ a b c "The Radio Logs of the USCG Itasca". TIGHAR. 5 June 2002.
  9. ^ "Amelia's Voice Heard by Amateur Radio Operator: Distress Signal Buoys Hopes for Ultimate Rescue of Aviatrix". Atchison Daily Globe. 4 July 1937.
  10. ^ Earhart 1937, p. 64
  11. ^ Earhart 1937, p. 70
  12. ^ a b Dorminey, Bruce (10 November 2014). "Aircraft Fragment Likely Not Amelia Earhart's, Says Investigator". Forbes. Retrieved 2018-05-27.
  13. ^ https://tighar.org/wiki/Nauru
  14. ^ https://tighar.org/smf/index.php?PHPSESSID=ngmi0j2ocn7p4q4c4scc8fknk5&action=dlattach;topic=1606.0;attach=8705;image 171 meters with 8-second period. Visible 15 nmi? Replaced in mid 1930s with fixed light. https://tighar.org/smf/index.php?topic=1606.15
  15. ^ "Twilight, Dawn, and Dusk". TimeAndDate.com. n.d. Retrieved 2018-05-27.
  16. ^ Bill Prymak, radio experts analyze Amelia Earhart’s bizarre radio behavior during her last flight By earharttruth on September 16, 2014, https://earharttruth.wordpress.com/2014/09/16/bill-prymak-radio-experts-analyze-amelia-earharts-bizarre-radio-behavior-during-her-last-flight/
  17. ^ https://tighar.org/wiki/Nauru
  18. ^ https://tighar.org/wiki/File:Flight-section-two.jpg
  19. ^ van Asten 2008, p. 25
  20. ^ addendum
  21. ^ van Asten 2008, p. 25
  22. ^ van Asten 2008, p. 25
  23. ^ Jameson 2016, p. 113
  24. ^ Jameson 2016, p. 114
  25. ^ Jameson 2016, p. 115
  26. ^ Jameson 2016, p. 116
  27. ^ Long & Long 1999, p. 229
  28. ^ Long & Long 1999, pp. 228–229
  29. ^ Long & Long 1999, pp. 235, 289
  30. ^ Longworth, R. C. (February 7, 1989). "Earhart Mystery May Have Run Out Of Gas". Chicago Tribune.
  31. ^ Nesbit, Roy (January 1989). "What Did Happen to Amelia Earhart?". Aeroplane Monthly. ISSN 0143-7240.
  32. ^ Gilmore, Daniel F. (June 18, 1982). "Did Amelia Earhart run out of gas?". UPI.
  33. ^ TIGHAR (November 1, 2011). "The Fuel System of NR16020". Earhart Project Research Bulletin. TIGHAR.
  34. ^ Long & Long 1999, p. 225. Notes on page 289 cite the search report of the Lexington.
  • Hoffman, Carl (3 April 2018). "Has Amelia Earhart Really Been Found?". Outside Online. Retrieved 21 May 2018. But the castaway theory is full of holes. Nikumaroro lies 350 nautical miles south of Howland, and Earhart herself reported that she was running out of fuel near the island. Those radio transmissions supposedly picked up by random people thousands of miles away? None have been verified as coming from Earhart. Navy search planes flew over Nikumaroro a week after her disappearance and saw nothing related to the aviator: not a human, not an airplane or the debris of one, not a smoke signal, not an SOS written with palm fronds. Also doubts Jantz and quotes Jantz.
  • Jameson, W. C. (2016). Amelia Earhart: Beyond the Grave. Lanham, Maryland: Taylor Trade Publishing. ISBN 9781589799905.
  • Long, Elgen M.; Long, Marie K. (4 November 1999). Amelia Earhart: The Mystery Solved. Simon & Schuster. ISBN 978-0684860053.
  • https://archive.org/stream/Amelia-Earhart/EHDOS#page/n49/mode/2up See page 51 for critique of Goerner.
    Also page 81: R. Kotsla, "Information on the Amelia Earhart Putnam Disappearance in 1937, Summary of"
    (1)(f) On 6 October 1943, CNI summarized the circumstances of Mrs. Putnam's disappearance as follows: "Gilberts were searched and British authorities there assisted in check up on all inhabited islands in vicinity. Mandates were not searched but Japs were cooperative (or appeared to be). Coast Guard cutter at destination (Howland Island) copied Earhart's transmissions and were convinced that she made emergency landing at sea within about 100 miles of Howland Baker. (Her last transmissions were tragic and near hysterical--impounded by Coast Guard Command at Honolulu). Studied conclusion at end of exhaustive search by Colorado, Swan, Itasca and Lexington planes (900,000 sq.miles) was that plane landed at sea and most probably went down with a rush and with crew inside. 20 kn headwind throughout flight from Lae and Earharts inability to "get a minimum" on her radio compass bearings of Itasca's transmissions appeared to be direct cause of her failure to reach island. Best data available from plane manufacturer convinced me that she did not have any margin of fuel whatever."
  • https://pacificwrecks.com/aircraft/electra/earhart/

Stable merge[edit]

Mark 14 torpedo[edit]

United States torpedo production during World War II
There should be a list of all the torpedo faults.
(Major) Torpedoes ran deep.
(Major) MI device did not work.
(Major) Contact exploder did not work. (Only appeared after MI disabled. Contact failures masked by MI prematures.)
Prematures
explode right after arming. Torpedo turning implies field change, so MI could fire.
explode just before hitting the ship. Probably MI detects hull and explodes instead of contact.
Torpedo bubble track does not imply should have hit.
myth? Or captain was right but torpedos ran too deep.
Duds
Circular runs
Designers put a generator beneath the core rod.
The seals leak. If a torpedo spent 15 minutes in the tube, it was done.
Did Chappell have a story about this? Tighten packing.
Testing.
There were a lot of tests.
Electric eye meant something else in the 1920s.
There should be section on BuOrd statistics.
Also Lockwood never provided (motivated) expert to BuOrd. Newpower (2006) makes a big deal of this, but would it have much effect? Lockwood's crew was very motivated. Did they dive on the cliffs and recover the dud? Yes. (Shireman 1998)
Storage: exploder uncocked but armed. I think this means no tension on the firing pin (balls not engaged) and armed puts the primer further away from the pin.
Firing: exploder cocked but unarmed. Tension on firing pin, but housing must spin to release it.
Can we estimate arming distance?
There should be a section about inability to produce torpedoes at a reasonable rate.
Number of submarines in the fleet. (S-boats don't count; they used shorter Mark 10s.)
6 Salmon-class submarine (1938)
10 Sargo-class submarine (1939)
12 Tambor-class submarine (1940-41)
Gato-class submarine 1941-43 showed up later
Balao-class submarine mid 1943
Number of torpedoes on hand at the start of the war.
Production rate. See chart.
about 1000 in first half of 1942; 166 per month.
Also compare response rate as Newpower pointed out.
Germans. MI, contact, depth. Early mission pointed to faults with torpedoes. Doenitz was swift. Problem solved in 6 months? Copy British design. Symbolic punishment of developers.
British. Investigate MI, but too iffy. UK just built a good torpedo. No faulty torpedo in service.
Japanese. Bigger is better. No faulty torpedo in service.
US. Early mission uncovered problem. Production problems meant nobody wanted to burn torps for testing until Lockwood had had enough.

Background[edit]

Torpedo depth control. Robert Whitehead in Fiume had trouble with depth control until it developed the "balance chamber" with pendulum. The balance chamber had water pressure push against a disk (cylinder?) that was balanced by a spring. "The inclusion of a pendulum stabilized the mechanism's feedback loop." Newpower (2006, p. 12) citing Edwyn Gray, The Devil's Device: Robert Whitehead and the History of the Torpedo (Annapolis MD, US Naval Institute, 1991 p33. This was sometime around 1870.

Whitehead had an improved exploder around 1877. Used a firing pin, but also had an arming mechanism driving by an impeller. (Newpower 2006, p. 15)

The US Navy tried developing its own torpedoes, but did not have much success. In 1891 E. W. Bliss Company signed a licensing deal with Whitehead to produce its Mark 1 torpedo. The US Navy started buying torpedoes from Bliss for $2000 each. (Newpower 2006, p. 15) It effectively ended the use of the Howell torpedo. (Newpower 2006, p. 16)

Around 1904, E. W. Bliss had its own torpedo that matched the Whitehead Mk. 5: the Bliss-Leavitt Mk. 1. That design had problems with roll stability, but Navy Lt. Gregory Davidson proposed using counter-rotating propellers. The proposal was adopted. (Newpower 2006, p. 18)

Navy decides to build its own torpedo factory. Newport breaks ground in 1907 and starts production in 1908. (Newpower 2006, p. 19) (But this has torpedo station producing Whitehead Mk. 5s!) 1912 saw the Bliss-Leavitt Mk. 7 as the last torpedo ordered by the Navy. Mk. 7 was the main torpedo of WWI.

p20 in 1911, Navy starts making its own contact exploder. Results in Mk 3 exploder in 1915. Also anti-circular run device.

Whitehead torpedos had a "war nose" contact exploder. Shear pin. Also impeller arming. (Wildenberg & Polmar 2010, p. 63)
There were earlier versions. Jolie (1978, fig 13 page 23) shows War Nose Mk. 1. Jolie (1978, p. 24) states War Nose Mk. 2 had whiskers that enabled an oblique hit to detonate the torpedo. Jolie claims war noses Mk. 3 and Mk. 4 never made it to production. The War Nose Mk. 5 was the first exploder "designed to fire on impact from any angle/direction", but it was designed for slow speed torpedoes and had trouble when torpedoes reached 30 knots.
Jolie, E. W. (15 September 1978). A Brief History of U.S. Navy Torpedo Development (PDF) (Report). NUSC Technical Document 5436.
Mark 3 exploder developed shortly before WWI and improved during. Wildenberg & Polmar 2010, p. 63
Terminology changed. War Noses were for torpedoes up to 1911. New designs starting 1911&ndask;1915 were termed exploders. The Exploder Mechanism Mk 1 had problems, so a Mk 2 followed, but Exploder Mk 3 improved on it. The Exploder Mk 3 was the first production exploder. (Jolie 1978, p. 24)
Contact exploder had anti-countermine device. Wildenberg & Polmar 2010, p. 63 Any explanation of how that worked?
While Mark 3 was being finished, BuOrd started considering magnetic influence. Wildenberg & Polmar 2010, p. 64
Project G53 funded with $25,000 in summer of 1922. Wildenberg & Polmar 2010, p. 64
Blister tests in 1924. 400 pounds of TNT on side of decommissioned USS South Carolina (BB-26). Also USS Washington (BB-47) during construction. Wildenberg & Polmar 2010, p. 64 Conclusion was impossible to cripple or disable a ship.
GE helped with G53 and ready for test in 1924. Captain Thomas Hart, officer in charge of the Torpedo Station, wanted to do a test, but he was not given a ship. Navy finally offered a ship in December 1925, and a test was done on 8 May 1926. Wildenberg & Polmar 2010, p. 64–65
There is a side photograph of the successful test. (Myers & Robinson 2018, p. 40)
There is a picture of Chester T. Minkler. (Myers & Robinson 2018, p. 21) (move elsewhere) "Mr. Minkler was one of the U.S. Navy’s leading experts in explosives. He served for 44 years at NTS Newport and established a solid reputation as “the father of the depth charge.”"
Chester T. Minkler is on the patent filed in 1932. Patent warns about high speed increasing sensitivity. Patent describes triode, but not thyratron. (Minkler 1946)
US 2397678, Minkler, Chester T., "Torpedo Exploding Mechanism", issued 2 April 1946 
"Chester T. Minkler, Devised Navy Bomb". New York Times: 50. 17 May 1972.
https://invention.si.edu/chester-t-minkler-papers-1911-1952
https://www.usnwcarchives.org/agents/people/134
https://www.usnwcarchives.org/repositories/2/resources/41
"10, 20, 40 Years Ago". Newport Daily News. Newport, RI: 8. 21 July 1955.
Myers, Bernard J.; Robinson, Joshua D. (2018). Milestones of Innovation: Naval Torpedo Station to Naval Undersea Warfare Center Since 1869 (PDF). Newport, RI: Naval Undersea Warfare Center.
Now here's some nonsense. "The Mark 6 was designed to sense the changes in the horizontal component of the earth's magnetic field caused by a ship as the torpedo approached the target." Wildenberg & Polmar 2010, p. 65
"The magnitude in the horizontal perturbation of the earth's magnetic field by a ship depends on the inclination of the earth's field to the horizontal. This inclination varies from zero at the magnetic equator to 90 degrees at the magnetic poles."
After the USS Indianapolis (CA-35) tests were done, the Mark 6 was finished. Although more live fire tests were wanted "... the Chief of Naval Operations would not permit the use of a live warhead." A destroyer was offered for tests, but there was a risk of holing her. The Torpedo Station did not accept the risk, so "no live fire test of the production version of the Mark 6 was ever conducted." Wildenberg & Polmar 2010, p. 66 (Torpedo station at fault.)
Where was the do not test with exercise head and countermand?
Where was the depth setting test issue raised? It does not make sense. Hydraulic ram will raise pressure on front of warhead (test port should read deep) and pressure port on after body should low (shallow). The readings would not correlate. There could be the issue of an exercise head. If the head is light, then they could both run shallow.
The fish net test was Lockwood's first, and it showed problems.
The chicken wire test was Lockwood's second, and it was successful.
Pendulum depth control.
O'Kane putting torpedoes into the mud.
Notice shape of Mark 14 warhead and Mark 15.

Start with Navy buying torpedoes from other vendors such as Whitehead and Bliss. Then it learns all it could from Bliss and starts doing its own production and development.(somewhere in Iron Men). That should be the Mark 10 production. TPS does not have the development or manufacturing expertise. NIH takes over.

The US Mark 10 is a reasonable torpedo, and the S-boats uses them effectively during WW II. (Newpower 2006, pp. 116–117)

Construction/design is by assemblies.

Exploder comes first.

Mark 3 exploder. Did it have counter-mining immunity? There was a friction issue that was solved with a stronger spring.

Where was Einstein and the crush head?

Worlds navies are recognizing that armor/blisters are becoming more effective against torpedoes. There are different responses.

U.S. believes TNT is good enough; later switch to Torpex; Wikipedia article says it was used starting in late 1942.

TPS also increased packing of warhead. US decides influence exploder is the trick (avoid the armor). Compare Germany (who developed MI and later copied British torpedo), UK (which did not oppose a military fleet (Britannia ruled the waves), so a contact exploder was good enough), and Japan (massive amounts of TNT).

G.B. doesn't want to redesign torpedoes so it looks for more powerful explosive; finds it in RDX. (MacDonald and Mack Partnership 1984)[page needed] G.B. also looks at influence exploder but decides it is unreliable and just uses contact exploder. G.B. has reliable contact exploder and reasonable weapon, but limited targets. Germany does not have significant shipping or risk much of its fleet. Compare Hood, Ark Royal, and Bismarck.

When did Torpex enter US service?

Germany develops new steam and electric torpedo and uses magnetic influence exploder. Doenitz goes nuts about depth control and enabling/disabling the MI exploder. Contact exploder is unreliable, too. Norway campaign is a disaster. German steals G.B.'s contact exploder.

Japan goes the more explosive is better route. Type 91 airborne torpedoes. (Newpower 2006, p. 113). Type 93 torpedo Long Lance has 490 kilograms (1,080 lb) and Type 95 torpedo has 893 pounds (405 kg). (Newpower 2006, p. 122) but another says 405kg model 1, 550kg model 2! Big explosive rather than the MI feature. Long Lance recognizes need for depth control. Japanese torpedoes are devastating: Pearl Harbor and Guadalcanal, but Japan does not deploy her submarines effectively;(Newpower 2006, p. 113) combat air cover does reasonable job of keeping torpedo bombers away.

Another tidbit: Long Lance was larger diameter. (where?)
It used different fuel. US declined development of Navol (hydrogen peroxide) torpedo. Such a torpedo does not need the space or weight of an air flask. Wildenberg & Polmar 2010, p. 67–70

Lockwood likes electric torpedoes.

Some German electrics land on US shores in 1942, TPS wants nothing to do with them. Westinghouse copies them (with GB contact exploder?).

In the summer of 1926, Lieutenant Ralph Waldo Christie became involved in the exploder's design.[1][2]

Should explain the nomenclature. The Mark 14 is the torpedo minus the war head, but the term is often used to refer to the whole torpedo.[citation needed] Internal to the Mark 14 were several assemblies. Cite to the pendulum depth keeping mechanism; it should be used on several other torpedo designs.

There's a war head that attaches onto the Mark 14. The warhead includes the explosive charge, but an exploder assembly attaches to the warhead.

The Mark 6 exploder bolts into the war head. There's a magnetic sense rod that then slips through the induction coil. The Mark 6 has an impeller wheel that is pushed by the flow of water. If drives a generator to supply power to the electronics; it also hoists the detonator into the war heads charge. If torpedo ran hot in the tube, no impeller motion, no arming of the torpedo, no charging of the capacitor.

Comment about the independence and interchangeability of the various Marks. Blessing and a curse. Balance issues.

In lieu of the warhead-exploder combination, there was an exercise head.

Both the warhead and the exercise head had a lead ballast on the bottom. Function as a keel.

  • Instructions for upkeep & operation of the mark VI mod. 1 exploder mechanism, Ordnance Pamphlet, Bureau of Ordnance, 1938, OCLC 51958048, OP 632
  • Torpedoes: Mark 14 and 23 Types, Ordnance Pamphlet, Bureau of Ordnance, 24 March 1945, OP 635

Assemblies[edit]

Page 15, OP 635, 24 March 1945 says

All assembled units and mechanisms in the torpedo are interchangeable in the Mark and Modifications to which they are assigned.

Torpedo body[edit]

Early torpedo: Mark 10 torpedo (this warhead was 500 pounds). Was warhead interchangeable?

Family of torpedoes:

  • Mark 13 (aerial, fat 22.5 in, stubby 161 in, slow 33.5 kn, 2216 lb, 600 lb (Mk 8 contact))
  • Mark 14 (sub, 21 in, 20 ft 6 in, 46 or 31 kn, 3280 lb, 643 lb (Mk 6))
  • Mark 15 (destroyer, 21 in, 288 in, 45 kn, 3841 lb, 825 lb (Mk 6))

Depth setting problem

Who described depth sensor being moved inboard to cooling flow?

Air flask supplies oxygen. Methanol for fuel. Drives turbine.

Two speeds.

Section floods with water to cool engine.

Gyro and depth are one subassembly, the Mark 12-3.[3] The same assembly is used on the Mark 18 torpedo.[4]

Heading control. Gyro.

Depth control[edit]

Depth control. Depth sensor and pendulum. Pendulum needed for stability. Consider way too deep, torpedo would just loop at depth. With pendulum, ailerons go neutral at reasonable attack angle.

Manual, pages 99–108. Uhlan principle. Page 98:

The diaphragm is acted upon by pressure of sea water while the torpedo is running. The diaphragm is held against this seawater pressure by the depth spring, connected to the lower depth-spring socket, plus atmospheric pressure in a pressure-tight air chamber. Since the pressure in the air chamber is constant for all practical purposes, the depth-spring loading controls the running depth of the torpedo. The pendulum is acted upon by gravity, and serves to limit the angle of inclination at which the torpedo may seek set depth, and at the same time tends to steady the torpedo at set depth.

The pendulum has another effect. When the torpedo is initially accelerated, the pendulum swings aft. Consequently, the torpedo thinks it is pitched up. The depth rudder will try to correct for that by ordering submerge. Consequently, the torpedoes go deep at first.[5]

Explain doubled instrument error. BuOrd "roll recorder".

Multiple faults: http://www.ibiblio.org/hyperwar/USN/Admin-Hist/BuOrd/BuOrd-6.html (Chapter 6)

Believe same mechanism used in the Mark 10 and possibly the Mark 18. There needs to be some adjustment for speed if the sensing positions are not the same. Can I find the sensing positions in diagrams. Or are there references that discuss the repositioning of some ports.

Is Bernoulli's principle explanation flawed?

Reynold's number[edit]

Calculate Reynold's number (stolen from that article).

The Reynolds number is defined as[6]

where:

1 gram per cubic centimetre (1,000 kg/m3)
21 inches (0.53 m)
45 knots (23 m/s)

From Viscosity article:

The dynamic viscosity of water is 8.90×10−4 Pa·s or 8.90×10−3 dyn·s/cm2 or 0.890 cP at about 25 °C.

Looks like 12 million.

Marks[edit]

According to late manual (pages 13–14)

  • Mark 14 out of service
  • Mark 14-1 out of service
  • Mark 14-2 out of service
  • Mark 14-3 out of service
  • Mark 14-1A two speed torpedo, gyro angles up to 90 deg left or right
  • Mark 14-3A like 14-1A, but gyro angles up to 160 deg. The 14-3A is a 14-3 with improved depth control
  • Mark 23 is a 14-3A with speed change mechanism removed and restriction valve locked in high power
  • Mark 23-1 is a Mark 23 with restriction valve drilled for high power

Generally, I think the A suffix is improved depth control (port moved from side to the interior).

Warhead[edit]

Manual, page 15:

The war heads provided for the Torpedoes Mk 14 and Mk 23 Types are the War Heads Mks 16, 16-1, and 16-4.

Manual, page 16:

War heads loaded with Torpex have a small charge of TNT in the nose of the head, and the War Heads Mk 16-1 and Mk 16-4 have a "topping charge" of TNT which fills the space between the after end of the Torpex charge and the after bulkhead of the war head. The War Head Mk 16, when loaded with Torpex, has a cavity in the after end of the charge in order to give the war head proper trim characteristics.

As I understand it, the original design used a charge was 500 pounds.

Some named person increased charge to 640 pounds or so. How that was done is debatable. One claim is the length of the warhead was slightly increased (could that be a reason the torpedoes did not fit in an S-boat?), but the exercise head was not changed.

Shireman states:

Adding insult to injury, earlier improvements by Commander King, although well-intentioned and initially successful, added to the depth-control riddle. When the additional 115 pounds of TNT were squeezed into the Mark XIV warhead, the exercise heads were not correspondingly altered to reflect the change. The extra explosive had been packed into the warhead by increasing density, so although the water-filled exercise head continued to occupy the same space as the warhead, it no longer had the same weight. Thus, the Bureau of Ordnance was using one version of the Mark XIV for testing and issuing quite a different Mark XIV.

That is not the complete answer. 500 + 115 = 615 pounds. Original warhead was TNT (1.654 g/cc), but Torpex (1.72 g/cc[7]) is better (but appeared later). There are some density issues, too. Shireman's comments are a little suspect. One cannot change the density of TNT. There are comments (I think in the Mark 14 manual) about placing extra TNT in a space that was previously void. See Manual, figure 4, page 15.

If the torpedo became nose heavy (but it was balanced, wasn't it?), then some up aileron is needed, so the torpedo runs deeper.

Calculation[edit]

The torpedo is 20 feet 6 inches. Why the 6 inches? Let it all be an addition to the warhead, so it would create increased volume. Assume the increased volume is a 20 inch diameter cylinder (10 inch radius):

That corresponds to 1885/231=8.16 gallons (8.159980918415). A gallon of water weighs 8.33 lbs, so 67.9728 lbs of water. TNT has a density of 1.654, so 112.4270112 pounds (116.913216 pounds of Torpex), close to the 115 pounds in some references.

Could expect the cavity to hold 500 pounds of TNT. That would be about 520 pounds of Torpex. Add 115 pounds for extra length to get 635. A 31 lb topping charge would give 666 pounds.

Overall length of war or exercise head is 47.28 inches (Mk 16). But summation of lengths gives 20.5 ft, so Mark 16 may refer to later, or there may have been a short warhead (for Mark 10?).

Capacities[edit]

Mark 14-3A torpedo used (manual, page 275). (16, 16-2, 16-3, 30, 30-2, 30-3 added to table.)

Mark type 14-3A total weight empty weight lead ballast liquid ballast/payload
16 warhead N ??? ??? ??? ??? When loaded with Torpex, there is a cavity to maintain trim (manual pg 16)
16-1 warhead Y 1053 264 ??? 666 lbs Torpex (said 66.6 lbs)
TNT topping charge
999info says pre/early war 507 lbs TNT; later 643 lbs Torpex
16-2 warhead N ??? ??? ??? ???
16-3 warhead N ??? ??? ??? ???
16-4 warhead Y 1057 264 ??? 600 lbs Torpex
TNT topping charge
999info says 666 lbs Torpex
30 exercise N ??? ??? ??? ???
30-1 exercise Y 843 429 287 414.5 (water);
end buoyancy 210
30-2 war exercise N? ??? ??? ??? ???
30-3 war exercise N? ??? ??? ??? ???
30-4 exercise Y 1053 503 367 550 (sp. gr. 1.35) (414.5*1.35=559.575)
Seems to recognized that 30-1 head is 210 pounds light; 135.5 lbs liquid ballast added; 74 lbs lead ballast;
end buoyancy 136

For war head, total weight may not include exploder. For simple exercise heads, total weight is empty weight plus liquid ballast.

Lead ballast (pull around) exists in warhead but warhead lead ballast not specified in appendix.[8]

999info is http://www.999info.net/Grunion/USN%20Weapons.pdf

Exercise head[edit]

There were several Marks. See appendix in the manual. These different heads probably give a clue about the warhead size changes.

Manual, page 21:

Exercise heads designated for the Torpedoes Mk 14 and Mk 23 Types are the Exercise Heads Mks 30, 30-1, 30-2, 30-3, and 30-4. The shells for the Heads Mks 30, 30-1, and 30-4 are drawn steel, with ogival nose sections and cylindrical after portions. The shell for the Exercise Heads Mk 30-2 and Mk 30-3 differs from the Heads Mk 30, 30-1, and 30-4, in that it is made of phosphor bronze, and is fitted for installation of an exploder mechanism and a device known as the "marker bomb", which will be described later.

There is a war exercise head that is used for .[9]

The war exercise head (Fig. 16) is used for experimental and instructional activities where the operation of the exploder mechanism is to be demonstrated. The head shell is fitted with an exploder mounting flange and casing similar to those in a war head, with the exception that the exploder casing is necessarily heavier because of the pressure built up in the head when it blows.

The Mark 30-3 is such a head (it is called out in Figure 16). Figure 17 suggests that the Mark 30-2 is also a war exercise head (marker bomb used in 30-2 would require a functioning exploder). The 30-4 is probably not a war exercise head because it is not of phosphor bronze. The only purpose of demonstrating the warhead would be to show of the MI feature. Not clear a torpedo would survive a contact on a steel hull.

O'Kane may describe exercise runs where torpedo is set to run underneath the target.

BuOrd pg 98:

Even during fleet exercises the error in depth could not be detected. To prevent impact damage to the weapon and target ship, practice shots were always set to run under the target. Concern over saving the torpedoes was so great that no one stopped to wonder just how far under the target the torpedoes were running.

Somewhere in Newpower (2006), at the tail end of the torpedo crisis, Blandy talks about a new exercise head.

British design[edit]

The British also needed to upgrade their torpedos, and they experimented with a magnetic influence exploder. The test results were unsatisfactory, so the resulting torpedo, the Mark VIII, had a simple but reliable contact exploder. There was no magnetic influence feature.[10]

The British were aware that capital ships were being armored against torpedos, so the explosive power of the World War I torpedo would no longer be adequate. That presented a design challenge: either increase the amount of explosive or find an explosive with more energy. Increasing the amount of explosive implied that existing torpedo tubes might be inadequate. Consequently, the British developed the more energetic explosive RDX.[11]

The British had a reliable torpedo when it entered the World War II. Perhaps the only disadvantage the British had was the limited number of targets for the torpedo. Brittain is an island nation that depends on shipping, so there were many targets for German submarines. German submarines attacked both British capital ships and merchant shipping. Germany's resources did not depend so much on shipping. Brittain had some success against the German invasion of Norway.[12] (There should be some comment about Brittain v. Japan in the Far East.)

The U.S. was behind on more powerful explosives. The U.S. military had decided that TNT was all the explosive it needed. The U.S. bet on the magnetic influence feature avoiding the armor, but then apparently retreated somewhat. There are stories of the U.S. adding another 116 pounds of explosive to the warhead (something that made the torpedo nose heavy and resulted in deeper running).[13] The British disclosed the existence of RDX to the U.S., and the U.S. started manufacturing it in 1940 (???). The availability of RDX led to the use of Torpex in U.S. torpedos.[14]

Japanese torpedoes[edit]

The Japanese developed several successful torpedoes. For the problem of ship armament, the Japanese just used more explosive — a lot of it. Cite warhead sizes. Newpower (2006) describes damage to U.S. Heavy Cruiser. Cruiser damaged but made it to port. Sent home for further repairs, but sinks on the way due to unrecognized structure damage caused by the torpedo.

German problems[edit]

The German torpedo of the time also had depth control problems due to using a light exercise head; the German torpedo ran 6.5 feet (2.0 m) deep.[15] The German torpedo designers knew the torpedo ran deep, but they ignored the problem because the magnetic influence exploder would make up for it. Some German officers were confined because they had let the problem persist.[16]

The German torpedo had problems with it magnetic influence feature; there were prematures.[15] The Norway campaign was an eye opener. There was some belief that ore deposits were providing fields that triggered the feature. Doenitz went back and forth about using the feature.[17] Ultimately, it was judged not reliable.

The German torpedo also had problems with its "clumsy" contact exploder.[15] The Germans captured some British Mark VIII torpedoes, which had a reliable contact exploder; the Germans simply copied its contact exploder.[18] Interestingly, some unexploded German electric torpedoes were later recovered from some soft, sandy, East Coast beaches. Westinghouse then copied the German electric torpedo rather than designing an entirely new one. The result was the Mark 18 torpedo, and it essentially used the British Mark VIII contact exploder design.[19]

Although the Germans had essentially the same problems with their torpedos, the Germans resolved those problems in about six months.[20] It would take the Americans a couple of years to work the bugs out of the Mark 14 torpedo.[21]

Reverse engineer the 30-4[edit]

The manual indicates that the Mark 30-1 and 30-4 are the same (long) dimensions, and some reverse engineering can be done. The Mark 30-1 is lighter than the Mark 16-4 warhead by 210 pounds, so it would not provide an accurate firing simulation. Making the exercise head heavier by 210 pounds would improve the simulation, but in order to recover the torpedo at the end of the run, it must have significant positive buoyancy. At end of run, the Mark 30-1 only has 210 pounds of buoyancy, so adding 210 pounds of lead ballast would not allow the torpedo to be recovered. From the tables in the appendix, the density of the liquid was increased to 1.35 (adding another 135.5 pounds) (volume is 100 pints? previous volume was 70 pints?). The lead ballast was increased by 74 pounds. (Pints of lead = 6.5255731922399.)

There is some evidence that Lockwood's fishnet test used a high density calcium chloride solution to match the warhead weight.[22][23] The test was probably done with a Mark 30 (short), but it is not clear only calcium chloride was used to increase the ballast.

Shireman:

Later in July, the Bureau of Ordnance responded to Lockwood's tests by announcing they were flawed and thus not conclusive. The Stateside bureau claimed improper trim conditions had been created when the field testers used an exercise head that was shorter than the warhead. Undaunted, Lockwood's team lengthened their exercise head to warhead length and immediately produced the same incriminating evidence.

At BuOrd, Commander James King tried Lockwood's test and confirmed Lockwood's results.[24]

Shireman:

Adding insult to injury, earlier improvements by Commander King, although well-intentioned and initially successful, added to the depth-control riddle. When the additional 115 pounds of TNT were squeezed into the Mark XIV warhead, the exercise heads were not correspondingly altered to reflect the change. The extra explosive had been packed into the warhead by increasing density, so although the water-filled exercise head continued to occupy the same space as the warhead, it no longer had the same weight. Thus, the Bureau of Ordnance was using one version of the Mark XIV for testing and issuing quite a different Mark XIV.
The problem of designing identical torpedo heads was solved by using Lockwood's calcium chloride solution, which correctly matched the warhead in size and density.

The above does not seem quite right; difficult to change density. Adding 6 inches to war head does make sense.

Manual is opaque about the preparation of the 1.35 specific gravity solution. Manual, page 121:

For information on liquid solutions, see OD 3837.

Mark 18 says use 1.355 specific gravity solution in specifications (Mark 18 page 4: "620 pounds of calcium chloride solution having a sp. gr. of 1.355"), but directions for exercise say fill with fresh water (Mark 18 page 153: "Fill head with fresh water").

There are some intriguing possibilities here. The initial depth tests may have been accurate with a Mark 16 warhead. The Mark 30-1 exercise head may be a lengthened (and heavier) Mark 30 head, and it may have shown the 3-foot depth error (manual corrections replace 10 feet with 13 feet). It may have required the extra 210 pounds of ballast to show the 10 foot depth error that Lockwood found. (Attempt to explain why BuOrd would claim 3 feet deep after Lockwood's fishnet test showed 13 feet.)

The Washington state torpedo station used sight depth tests first; later it used fishnet tests.

Cutaway exercise head[edit]

File:Mark XIV torpedo warhead cutaway.JPG

Hydrostatic pressure port position[edit]

Mark 6 exploder[edit]

Early version of the Mark 6 exploder
Mark 6 Mod 1
Mechanical drawing of the Mark 6 Mod 1 exploder.
Mark 6 Mod 5 exploder showing two positions of the detonator and safety chamber. This exploder replaced the mechanical firing pin of the Mod 1 with electric detonator. Consequently, it does not have the fingers, the firing ring, and other mechanical parts of the Mod 1, but the base still has the flanges where the fingers would attach. Strangely, the solenoid is still called out. Why? It would not be needed because the thyratron could fire the electric detonator.
Main article: Mark 6 exploder

German magnetic influence mine. Germans didn't use it on a torpedo.

Contact exploder. The safety elevator made the firing pin motion perpendicular to travel. Would work at high angles or low speeds. Early Mark 6 had a complicated inertial ring mechanism; later was a simpler ball switch.[25]

It is not clear that the inertial ring mechanism was flawed. The problem that Lockwood uncovered was with the firing pin mechanism undergoing severe deceleration. The same firing pin was used for the magnetic influence trigger, but the torpedo is not decelerating at instant of triggering. The Mark 6 exploder still used an electronic circuit to trigger the detonation (but it may have bypassed the thyraton). It would be interesting to find out if the thyraton failed under high deceleration.

Shireman points out that BuOrd had already seen this problem in the 1930s. BuOrd tests found the firing pin was unreliable, but it solved the problem by using a more powerful spring to overcome the forces. Unfortunately, the spring worked at 30 knots but not at 46 knots:

During the 1930s, the Bureau of Ordnance had conducted similar tests designed to ensure a reliable contact mechanism in time of war. The Newport Torpedo Station flung torpedoes against steel plates over sand and discovered then that the firing pins failed to strike the caps with sufficient force. Their solution was to increase the strength of the firing spring. The tighter spring seemed to solve the problem, but it did so at the speed of 1930s torpedoes. Torpedo speeds had increased to 46 knots by World War II, and this increase created greater impact forces. The increased speed essentially negated the strengthened spring. If Tinosa's torpedoes had been set for slower speeds or obtuse angles, Tonan Maru No. 3 would not have escaped. It took almost two years of wartime trials and tribulations, but American submariners were finally equipped with reliable and effective torpedoes.

It is not clear if BuOrd did the tests at full speed. When did the Mark 14 have its high speed? Or were the tests done for lower speed torpedoes?


After the torpedo is launched, sea water travels through an open channel on the bottom of the torpedo. The water hits an impeller that turns a shaft connected to a generator. The generator will supply power to the exploder. There is a worm on that shaft, and it engages the worm wheel that spins a vertical shaft. There is another worm on that shaft (not in the schematic but on the other picture), and it turns the delay device gear. Initially (A), a switch touching that gear grounds the generator field winding and prevents the generator from producing any power. (That should make the the generator shaft difficult to turn.) Notice that some teeth are missing on the gear at the 11 o'clock position; there's also a hole in the gear at the 3 o'clock position. This gear will make almost a complete turn and reach (B). At that point, the hole is underneath the switch and causes the switch to stop shorting the field winding. The generator starts producing electricity to power the exploder. At the same time, the missing teeth are now at 9 o'clock, so the worm no longer engages the delay gear; the delay gear stops moving.[26]

The vertical shaft continues to turn and has been doing other work. Through a gear train, it causes the arming gear to rotate. That causes the safety chamber to rotate and has the detonator holder rise out of the safety chamber. When the detonator is inside the safety chamber, it cannot detonate the tetryl booster and keeps the weapon safe. Once it rises out of the safety chamber, the booster can be detonated.

With the early Mark 6, something else is also going on. Initially, the threads on the arming screw prevent the screw from moving up and keeps the trigger plate from rising. As the arming gear rotates, it also raises the arming screw. Raising the arming screw further compresses the firing spring. Eventually, the arming screw runs out of screw thread, so it stops just above the arming gear but below the firing pin. The arming screw still retains the ball(s) that holds the firing pin in place. When the torpedo hits something, the firmly attached exploder body suddenly decelerates, but the firing ring has inertia and keeps moving. The trigger plate moves upward and raises the arming screw high enough so the ball(s) leaves the groove in the firing pin; the firing pin is then driven by the firing spring and strikes the detonator.

In the early Mark 6, detonation is even more convoluted for the magnetic influence feature. A thyratron tube fires and energizes the solenoid which moves the armature. The solenoid does not have enough energy to cause detonation, so it just moves the pawl so it will engage the spinning ratchet that is also on the vertical shaft. The ratchet moves the pawl which moves the pawl arm. The pawl arm then moves the firing ring which initiates the same sequence as above.[27]

The later Mark 6 did not use a mechanical detonator. It still used the arming gear to raise the charge out of the safety chamber and the delay device on the generator, but a simpler electric detonator was used.

MI[edit]

Many belived in MI, and the feature did work at least once in awhile. Wahoo down the throat shot.[28]

Torpedo scarcity drove desire to use MI. There's some strangeness here because most of U.S. submarine attacks were against merchantment that had no armor. Tang was sinking merchantment with single torpedos late in the war, so it was not required to do the under the keel shot.

Magnetic influence exploder. Disable by removing core rod. Does anyone discuss degaussing of ships and impact on torpedo? Newpower 2006 does.

Interpreting the bubble track.

Running deep would take torpedo out of the influence of the ship.

Who described there were two tests and one failed. That was a torpedo article in several parts. I need to find them. (all at archive.org?)

If the MI feature were designed so that the war head would detonate when it was 10 feet under the keel, then it could also detonate when it is 10 feet away from the hull. That's probably why the MI feature needed to be disabled to avoid premature detonation. If the ship were not degaussed, then prematures might be more likely.

Higher torpedo speed would make the MI more sensitive. The earth's magnetic field is static; a ship's magnetic field would also be static. The induction coil senses a change in the field. If the relative speed to the target is 50% faster, then the signal is 50% larger. Was the MI designed for low-speed torpedoes and then used on high speed torpedoes?

BuOrd 100–101:

So attractive was the goal that in the immediate postwar years the Bureau experimented with a variety of ways to produce an influence explosion. Sound controlled torpedoes, a water kite above a deep running torpedo, and the creation around a torpedo of an electric-magnetic field that would be disturbed by the entrance of a metallic body such as a ship were all tried, then abandoned. But the idea of using magnetic influence opened a new field and on June 30, 1922, the Bureau of Ordnance instituted at the Newport Station the "G-53 Project5" that eventually produced the Mark 6 exploder. The project seemed especially timely since the same y;ear witnessed the adoption by most nations of new antitorpedo structural protection for their first line fighting vessels. More than ever, underbottom explosions seemed desirable, and the G-53 Project was allowed to suffer less from budget restrictions than the testing and development of other weapons in the arsenal of underwater ordnance.
With the help of the General Electric Co., which produced the generator and developed thyratron electronic tubes for the project, Newport had the Mark 6 ready for testing by 1926. Although the idea of employing magnetism was first crystallized by Germany, the United States' exploder represented a new line of development. The swinging of a compass needle when approached by the magnetic mass of a ship activated the German device; our mechanism utilized the variations in the intensity and direction of the earth's magnetic field adjacent to the hull of a vessel to actuate the mechanical pistol that set off the explosive charge. On May 8, 1926, 4 years of work were crowned by success. A submarine hulk was towed to sea, then sunk by the first shot of a torpedo equipped with the new magnetic influence exploder.

Telling comment on page 101:

Never again during the 19 years of prewar exploder development was a destructive test made with a torpedo equipped with a magnetic influence exploder.

BuOrd page 102 about prematures:

Their testing was inadequate. Evidence of that fact came in 1939, when Newport reported to the Bureau that the exploder was giving unexplained prematures. Admiral Furlong arranged for a physicist to visit the station and investigate the failures. For approximately a week, the scientist and his assistants worked with the device. Four sources of prematures were uncovered. Even more significant, the investigator reported to the Bureau that the responsible engineers at Newport were not employing proper tests on the Mark 6. Corrective steps were ordered by the Chief, but subsequent events proved that the remedial action, like the original tests, was inadequate.

Contact feature[edit]

The original contact feature was an inertial ring. (There's a reference for that.) At contact, the ring was supposed to slip out of place and release the firing pin. A spring would drive the firing pin into a primer, and that primer would initiate the tetryl explosive booster charge that sat on top of the detonator housing. (Loose fitting is described in the manual procedures for assembling the exploder to the war head.)

A safety feature is detonator sat in a safety housing, and the primer could initiate the booster charge without setting off the main charge in the war head. When the torpedo started its run, an impeller would spin and raise the detonator (and the booster charge) into a well of the warhead. The more confined space would allow enough of a shock to detonate the warhead's main charge.

A further safety feature may have been the raising of the detonator mechanism also cocked the firing pin spring.

The magnetic influence feature also needed to detonate the main charge. That was accomplished with an induction coil sensing a change in the magnetic field and producing a voltage. That voltage ultimately triggered a thyratron that energized a solenoid. The solenoid would trip the firing pin.

The problem with the setup was the firing pin moved perpendicular to the torpedo's travel. At impact, there would be high deceleration forces that would interfere with the firing pin movement (and even distort the pin). Consequently, the BuOrd replaced the spring, firing pin, and primer with an electrical detonator (blasting cap). For contact firing, a new sensor was needed to fire the electrical detonator, and that sensor was the ball switch.

Comments about leaving many other things the same.

Air flask destroyed on impact, spray of water against the ship, but not high order detonation.

There were some crazies here. Perhaps Einstein consulted on the Mark 3 exploder problem. Failure at higher torpedo speeds. Result was stiffer spring.

Little is said about the Mark 5 exploder (Mark 6 without the MI that was issued to most submarines).

Many claimed duds. Daspit final straw. Fire into HI cliffs. Third was a dud. then the drop tests.

Lockwood's tests[edit]

Lockwood ordered some famous tests, but the some details are not clear.

The fishnet test. At this point, it's not clear that the exercise heads were updated. One source suggests the use of a denser solution, but it is not clear who is responsible for that. BuOrd said it ran its own tests or calculations, and admitted 3 feet. If BuOrd ran fishnet tests, then why didn't it come up with 11 feet as did Lockwood? Did BuOrd use a light warhead with fishnet? Or did it use a compromised depth gauge? Why would BuOrd come down on Lockwood so hard? The fishnet experiment is pretty basic. Fishnet could deflect before ripping, but 8 feet? Any information about how the fishnet was anchored above and below?

Bluff firing. Real warheads, so hard to dispute conditions. The some wag comment about going under the island.[29]

The drop tests seem most poorly reported. Filling the warhead with concrete just seems silly, but that is reported in Newpower (2006) by a guy who was there. Dropping entire torpedoes just to test the warhead also seems silly; why destroy a expensive Mark 14? Dropping a head full of explosive seems a bit dangerous (what happens when it is a dud?). The test must be sure the primer actually fires. Operation Pacific shows dropping just the warhead, but the exploder shape is wrong. Also, the real test would require arming the torpedo just before he drop. If he used real warheads, then removing the explosive would be troublesome. If he had a supply of right exercise heads, then he could use those (the simplest route). If the explosive were absent, then what was used as a filling? Concrete seems more trouble that it is worth. An exercise head is easily filled with liquid. An emptied warhead could be filled with a small amount of sand (which is much denser than the explosive); there is no reason to have the inert material take a set. The concrete would make sense as a replacement for Mark 14; much cheaper and reusable.

Result: 7 out of 10 failed at perpendicular.

Dropping just the warhead without an equivalent Mark 14 mass behind it could show a higher acceleration because the nose does not crush as much.

The drop from 90 ft. d = 90 feet = 0.5 a t2. Acceleration of gravity (a) is 32 ft/sec/sec, so t = sqrt(90/16) = 2.3. Thus v = 2.3 * 32 = 76 feet per second (45 kn).

About the same time as the drop test, there was a chicken wire test. Some skippers believed the torpedo was still running deep. Lockwood's tests showed the depth was about right.[30]

Prematures[edit]

Firing geometry includes reach

The explanations here are confused and varied. The prematures hid the failures of the contact exploder.

Arming distance. If the torpedo armed before settling down on a course, then the MI might fire.[31]

Gaskets leak. A real shock. If torpedo was immersed for 15 minutes, it got water logged. There were many discussions about gaskets and whether sub personnel did the seals right. Anecdotal story about Gallatin believing impeller packing was not tight enough. Mechanism is salt water causes a short that triggers exploder as soon as it arms.

Infamous BuOrd draft letter. Degaussed ships may not trigger MI. Ungaussed ships may trigger MI far away. Latitude makes a difference. The close aboard detonations.

Robert F. Marble's ten questions[edit]

Mark 14-3A Torpedo and its MK 6 Exploder

I have ten torpedo related questions for you.........

  1. What is the arming distance of the exploder mechanism on a MK 14 or MK 15 torpedo?
    350 yds (MK 14 manual, page 20). 350 yards (320 m) and 46 knots (24 m/s), so about 15 seconds.
  2. Why are the warheads made of phosphor bronze on MK 14 and MK 15 torpedoes and steel on the MK 13, MK 18 and MK 23 torpedoes?
    Phosphor bronze is nonmagnetic and will not interfere with the MI feature. (Should be in MK 14 manual.)
  3. What type of exploder mechanism is installed in a MK 23 torpedo?
    Mark 6 Mod 5. MK 23 is a MK 14 with one speed and no MI. (MK 14 manual page 17; fired by direct impact.)
  4. Can a MK 15 (surface to surface) torpedo be used in a submarine?
    Diameter is OK, but I think it is too long (+42 inches) for the torpedo tube or perhaps loading into the submarine. It also may not have the control setting features of the Mark 14.
  5. What prevents a torpedo from rolling over after it is launched?
    Lead ballast along the bottom of the exploder or warhead. (Should be in MK 14 manual.)
  6. How does the anti-countermining feature on a MK 6 exploder mechanism work?
    I haven't seen a description of this anywhere. the MK 3 exploder developed the technology. I suspect all the magic is in the detonator ring, but I don't see how it would work. Vertical forces would not affect the ring, but they would not be dominant in a nearby detonation. BuOrd later concluded that the anti-countermining feature did not work. I think Rowland and Boyd 1953.
  7. How does the anti-circular run feature work on a MK 6 exploder mechanism?
    I don't think there is such a beast on the MK 6 exploder. The likely place for circular run detection is in the gyro control unit. If the gyro travel is too far, then that unit could order the torpedo to dive because it also controls depth. From what O'Kane said (Wahoo IIRC; he was in USS Argonaut (SM-1) off Midway when the war started; sonar shots from deep submergence; a submarine tactic was to have torpedo circle overhead, but I don't know how such a torpedo would be fired), I don't think there is an anti-circular run detector on the Mark 14 torpedo; there may be on the MK 15. Later versions of the Mark 14 would allow for 160 degree turns onto the track.
  8. What is the maximum depth for opening a fleet submarine's torpedo tube for launching a torpedo?
    Probably the seal depth for the torpedo, but it may be an air issue to eject the torpedo from the tube.
  9. What successful post-WWII modifications were made to the troublesome MK 6 exploder mechanism?
    See Rowland and Boyd 1953.
  10. What is the difference between a run-down gyroscope and a constant-spin type?
    The run-down gyro is spun up in the torpedo tube and then just coasts (runs down) after launch; the constant-spin gyro uses the air flask to keep it spinning during the entire run.

Edits before revert[edit]

References[edit]

  1. ^ Blair 1975, pp. 54–55, 304 https://books.google.com/books?id=KGjfqe6DcrEC&pg=PA54&lpg=PA54&hl=en&f=false.
  2. ^ Newpower 2006, p. 27 has Christie at Annapolis for post graduate studies in ordnance engineering starting December 1920; after finishing at Annapolis, he went to MIT and graduated with a masters degree in June 1923. In the summer of 1926, he reports to Newport and works on successor to Mk 3 exploder. Newpower 2006, p. 29, by 1938 Christie was Head of the Torpedo Section at BuOrd and involved in providing Mk 6 and Mk 14 torpedo to the fleet.
  3. ^ Manual page 77
  4. ^ Mark 18 manual, OP 946, page 116
  5. ^ Manual [page needed]
  6. ^ Cite error: The named reference Sommerfeld 1908 was invoked but never defined (see the help page).
  7. ^ OP 1664 Vol I, page 3. http://www.hnsa.org/doc/ordnance/. There was a Torpex 1 and a Torpex 2.
  8. ^ Manual pages 15, 275; density Pb 11.34
  9. ^ Manual page 24–25
  10. ^ Newpower 2006
  11. ^ MacDonald and Mack Partnership
  12. ^ Newpower 2006
  13. ^ Disparate
  14. ^ When Torpex appeared not clear. Is it the same formulation as the British used in their torpedoes?
  15. ^ a b c Patrick 2012 stating, "Many U-boat commanders suspected that torpedoes were running too deep. The Torpedo Directorate had known for some time that torpedoes were running six-and-a-half feet deeper than intended because the exercise heads used for calibration were more buoyant than warheads."
  16. ^ Newpower 2006, page needed.
  17. ^ Newpower 2006
  18. ^ Newpower 2006
  19. ^ Newpower 2006
  20. ^ Newpower 2006
  21. ^ Newpower 2006
  22. ^ Blair, p 276
  23. ^ http://www.military.com/Content/MoreContent?file=PRtorpedo
  24. ^ Shireman
  25. ^ Milford, Part I, footnote 30. Mark 6 Mod 5 has the ball switch.
  26. ^ This operation is explained in chapter 12 around page 275–276.
  27. ^ Explained early in Newpower 2006.
  28. ^ Newpower 2006
  29. ^ Believe it is Submarine Commander: A Story of World War II and Korea, Paul R. Schratz, page 84. http://books.google.com/books?id=wHhpv6YgYgUC&pg=PA84&lpg=PA84&source=bl&hl=en&sa=X&ved=0CCsQ6AEwAA
  30. ^ Newpower 2006
  31. ^ Newpower 2006

Further reading[edit]

indiscrimant links to resolve[edit]

Foundry link[edit]

Subsequent notes[edit]

Looking for details. History of Mark 3 exploder? Christie was only a lieutenant when working on Mark 6; who was in charge? Mark 10 evolution.

Where is the targeting advice letter? Newpower, Chapter 8, Note 80.

Newpower also has Blandy's early claims to performance and subsequent performance. Subsequent performance showed that shallow running torpedoes would prematurely explode about 9 percent of the time.

Also, the test firing data of the Mark 10 torpedo with the MI head involving USS Indianapolis (CA-35) were apparently lost. Southern latitudes were not good for the MI. Newpower ch 8.

xxx

Battle of the Atlantic[edit]

Somewhere there were details about the May 1940 introduction of Kennbuch procedures and then a fast break. I think it was Tony Sales' pages, and I probably put it in the talk page timeline.

There were several pinches. Sales' http://www.codesandciphers.org.uk/virtualbp/navenigma/navenig6.htm

Seizing the Enigma has the milchcow attack that went awry.

Enigma[edit]

Tony Sale's site has lots of info, http://www.codesandciphers.org.uk/enigma/index.htm

M4 project, http://users.telenet.be/d.rijmenants/en/m4project.htm

Enigma Breaking Project has details, messages, German code forms, http://www.enigma.hoerenberg.com/index.php?cat=Welcome

Steven Hosgood gives details about Banburismus, http://stoneship.org.uk/~steve/banburismus.html

Issues[edit]

There are several issues to understand here.

Bombe menus.

The doubled indicator procedure could find a stop. If it were right, then the wheel order and start position were known, but the ring position and steckers were not.

If someone started decrypting the message, it would turn to gibberish at the turnover. That could give the ring setting for the fast rotor. I cannot find the ring setting for the middle rotor unless I have a turnover on that rotor. Same for the slow rotor. A few hundred characters should give me the middle rotor, and several thousand characters the slow rotor.

Sales suggests the different turnover positions would give away the rotor. If I know the message indicator is zzA, and I can somehow detect a turnover at zzR (Royal), then I know the rotor is I. But how do I detect the turnover?

The ring setting and the plugboard protect a message indicator sent in the clear. Without rings or plugboard, then one would just try all the rotor orders at the message indicator setting.

The plugboard is a monoalphabetic substitution. Is the M4 project hillclimbing just for calculating that with mono-, bi-, and tri-grams?

Message numbering and forms[edit]

The German forms.

Closely run affair[edit]

I think the Herivel tip article leads to a comment by Alexander about its value. The implication is Germans added complexity, but the complexity was incrementally doable. If there had been some major changes, then they would have defeated the attacks.

The Polish bomby relied on message indicator repetition. That system was working fine, but Turing thought it wouldn't last. The British bombe was designed to look for cribs, and it was already being designed/built (check) when the indicator repetition stopped.

Sales has the pinches as necessary because the tables could not be figured out.

The HT may be even more clever. Guess rotor order, but the tip gives both ring settings and rotor positions.

Various contact protection articles[edit]

Company had nice pictures, but made some dubious claims.

Audion[edit]

Want frequency characteristics. GBW product.

Blue glow. Can gas breakdown law infer a level of vacuum in the audion? (May not be a glow discharge?)

Ultraaudion, audio converter, adjust filament power, "hissing", imperfect vacuum, shot noise oscillator "works" like Poulsen arc, "siren effect".

The Quarterly Journal of the University of North Dakota, Volume 5, 1915, http://books.google.com/books?pg=PA198&lpg=PA198&dq=frequency+characteristics+audion&sig=sokxdnjRiHXt0hXN6C0j4nosgIM&ei=Ewe6UfPvEcb_igKDoYCQAQ&id=kN8hAAAAMAAJ&ots=P1RqY-boCa&output=text

Poulsen arc[edit]

Several early radio publications have explanation. NBS explanation of three modes, only one important.

Nobody gives a good explanation of the IV characteristic. Unidirectional. Carbon electrode can get hot and have thermionic emission. Metal electrode is cooled, so poor thermionic emission. Magnetic blowout.

Ion issue. Looking for gas breakdown to arc but then ability to quench the arc?

IRE Vol 5. P. O. Pedersen, "On the Poulsen Arc and its Theory" date=August 1917 volume=5 issue=4 page 255

At Hathi Trust? http://catalog.hathitrust.org/Record/008616679

Vol 5 also has Armstrong, A Study of Heterodyne Amplification by the Electron Relay, pp 145

Yagi on transformers, page 433

Glow discharge simulation[edit]

Gas discharge tube[edit]

Electric glow discharge

The typical GDT has a low pressure noble gas. For the present, assume that the electrodes do not emit any electrons. Also assume there are no ions.

Consequently, the GDT is an insulator; no current flows.

  • Lieberman, Michael A.; Lichtenberg, Allan J. (1994), Principles of Plasma Discharges and Materials Processing, New York: John Wiley & Sons, p. 453, ISBN 0-471-00577-0, Considering current density as the controlling variable, the voltage-current characteristic of a dc discharge is shown in Fig. 14.2. The flat region with slightly negative slope dV/dI is that of the normal glow. From low currents, the region below IA is called a dark or Townsend discharge. The glow gradually builds up until a transition is reached, with hysteresis, entering the normal glow at a voltage VS. The voltage remains constant as the current increases until IB, at which point there is an increasing voltage-current characteristic called the abnormal glow. A further increase in current resutls in a rather abrupt transistion at IC, again characterized by hysteresis, to a considerable lower voltage discharge known as an arc discharge.}}

Dark current[edit]

There will be some ions created by light, cosmic rays, or radio activity. Consequently, there will be some positive ions and electrons. With no applied voltage, there is no significant drift and they recombine. Applying a voltage causes the heavy positive ions to drift toward the cathode and the light electrons to drift to the anode. Some could still recombine.

Increasing the applied voltage causes all the generated ions to be swept to the anode and cathode before they recombine. Consequently the current reaches a saturation level of IS that stays constant as the voltage is increased.

See Rutherford (1899, pp. 112–113).

Glow discharge[edit]

At some point, the drifting ions gather enough energy that they can generate more carriers. This region is Townsend discharge, and there are two dominant mechanisms. The first mechanism is electrons colliding with neutral atoms to generate an additional positive ion and electron. The positive ion drifts to the cathode, and the negative electron drifts to the anode. Avalanche multiplication is happening. Now there are two electrons, and each of them may gain enough energy to kick loose more electrons.

Discuss photons... Start of glow.

The second mechanism is secondary emission. When a positive ion crashes into the cathode, it may liberate another electron that will then drift to the anode (and, possibly, free more carriers).

Breakdown[edit]

Uncontrolled generation.

Switch modes.

Suggestion that secondary emission is now 1 for 1 at ignition / strike.

Extinction voltage.

GE reference says negative resistance, but Tube reference does not.

Normal and abnormal glow[edit]

Start of thermionic emission[edit]

Local heating of the electrodes makes it easier to develop a current.

Arc discharge[edit]

Arc converter
Three modes; switching is Poulsen
Bureau of Standards (24 May 1921), The Principles Underlying Radio Communication (Second ed.), U.S. Army Signal Corps, ISBN 9781440078590, Radio Communications Pamphlet No. 40 http://www.forgottenbooks.org. See following page 400. FSK keying page 415.
Arc lamp
Avalanche breakdown
Electric arc
Electric discharge in gases
Spark gap
Vacuum arc
Several articles related contact damage and protection
Potter and Brumfield, spike and crater, http://relays.te.com/appnotes/app_pdfs/13c3203.pdf

Voltage becomes similar to ionization potential.

Arc discharge requires electrode emission. Thermionic or high field.

Self destruction.

Negative resistance here.

Glow character[edit]

An electric glow discharge tube featuring its most important characteristics: (a) An anode and cathode at each end (b) Aston Dark Space(c) Cathode glow (d) Cathode dark space (also called Crookes dark space, or Hittorf dark space) (e) Negative glow (f) Faraday space (g) Positive column (h) Anode glow (i) Anode dark space.

Cathode dark space.

Striations.

Faraday dark space.

Column.

Anode glow.

Dubious references[edit]

  • Millman, Jacob (1958), "12 Electrical Discharges in Gases", Vacuum-tube and Semiconductor Electronics, McGraw-Hill Electrical and Electronic Engineering Series, McGraw-Hill

Has lots of detail, but does not show reorganization of gas. Has negative resistance region at Townsend breakdown; acute angle.

  • Ott

Ott may have referred to glow region as Townsend discharge. Has an arc being a metal-arc discharge and suggesting that metal ions are involved.

Neon spectra[edit]

Stuff[edit]

I disagree with the notion that devices such as a gas discharge tube exhibit negative resistance.

Consider a non conducting gas tube just below the strike voltage. Now increase by one volt to fire the tube. If the terminal voltage is held constant, then the current increases dramatically instead of falling (as it would in a negative resistance).

If the terminal voltage is then decreased, then the current also decreases. Decrease the voltage enough, and the tube quenches. Where is the negative resistance? I see a state change and hysterisis due to that change, but I don't see a negative resistance.

The I-V characteristics are complicated. Start out with Townsend discharge that has avalanche multiplication at high end. Then there is dark discharge, glow discharge, and arc discharge. Herbert J. Reich. Noise and shielding book (which is interested in supressing arc to minimize contact damage) only does strike / glow model.
Neon lamp#Description has an I/V characteristic graph with a negative resistance region, but it is unsourced. Some technical articles suggest state change.

In contrast, if the same terminal measurements are made varying the voltage across a tunnel diode, there's a region where the voltage is increased but the current decreases.

If the independent variable is the current, then yes, there's a point where the current increases but the terminal voltage falls because the tube finally strikes. That transition, however, is also a discontinuous jump on the device IV curve. With the jump, there's no clear notion of a small signal impedance.

I don't see a nonlinearity or hysterisis implying a negative resistance.

Where's a reference that states a gas discharge tube is a negative resistance?

Does a circuit that can be broken somewhere in the feedback loop and display a negative resistance qualify? Negative resistance should not be just another word for power gain. That is the sense used in microwave oscillators, but the wizard behind the curtain is an amplifier.

Philbrick ref (page 8) says it resembles a Howland with infinite load.

Arcing[edit]

In addition to Ott...

History of metal arcs. 1903; rediscovered. Voltage less than 6 volts to avoid breakdown.

  • Holm, Ragnar (1967), "Bridge transfer and short arc transfer at contact separation", Electric Contacts: Theory and Application, Berlin: Springer, pp. 338–347, doi:10.1007/978-3-662-06688-1_65, ISBN 978-3-662-06688-1
    Abstract: Bridges in opening metal contacts have been observed a long time ago. Sundorph [1], in 1903, and particularly Angelica Szekely [1], in 1924, studied such bridges. The latter worker stated that the material of the bridge was derived from the anode, and that the bridge always attained such dimensions that its voltage finally remained constant, of the order of 0.5 to 1 V. Visible bridges were easily obtained between iron electrodes. Her publication was little observed. In 1938, the bridges were rediscovered and studied by Betteridge and Laird [1]. They succeeded in photographing stabilized molten bridges between platinum contacts. They showed that the final bridge voltage, before rupture, was of the order of 0.7 to 2 V, depending on the contact metal. They also observed that the hottest section is not in the midst of the bridge. They assumed the Thomson effect to be responsible for the displacement. However, R. Holm [29] remarked that the Thomson effect, at that time measured on solid metals only, had the wrong direction for an explanation of the displacement in platinum. In the following, the Thomson effect can be accepted as the cause of the displacement in platinum because Llewellyn Jones has shown that with platinum contacts the effect changes its sign when the metal melts; see Fig. (66.02).
  • Kharin, Stanislav (November 1997), "Mathematical model of the short arc phenomena at the initial stage", Electrical Contacts, Proceedings of the Forty-Third IEEE Holm Conference on Electrical Contacts, IEEE, doi:10.1109/HOLM.1997.638055, S2CID 42566186
    Abstract: A mathematical model is presented describing transient phenomena accompanied by a vacuum short arc at the initial stage of contact opening. It enables one to describe the evolution of transient short anode dominated arc, which appears just after the rupture of molten bridge, into diffusive cathode dominated arc. The model of an vacuum arc includes equations for the cathode region (sheath, ionization zone, cathode surface and bulk), arc column and anode region. The heat flux components entering and leaving electrode surfaces owing to ion bombardment, electron emission, inverse electron flux and radiation from the arc, melting, evaporation and heat conduction are taken into account. The peculiarity of this model is the radial dependence of unknowns values and anode spot extension with due regard to both arc column expansion and increasing of conic arc base during contact opening. The model enables one to find the temperature and electrical fields, heat flux components, electron, ion, and inverse back-diffusion components of current density in dependence on current, opening velocity, contact gap, and properties of contact materials. It was found also that the duration of transient anode dominated arc depends on the parameters of liquid bridge. Transition of transient anode arc into diffusive cathode arc is formulated. The results of the calculations for the copper and molybdenum contacts are given for the current range from 100 A to 500 A and opening velocities from 0.5 m/sec to 20 m/sec
    https://www.researchgate.net/publication/3723837_Mathematical_model_of_the_short_arc_phenomena_at_the_initial_stage
  • Haase, Roman (February 2015), "Simulation of relay contact bouncing including a short arc model", Electrical Contacts, Proceedings of the Annual Holm Conference on Electrical Contacts 2015, IEEE, doi:10.1109/HOLM.2014.7031064, S2CID 22590940
    Abstract: The continuous growing requirements on power relays make it necessary to handle higher switching capacities which are represented in an increasing number of switching operations resulting in a demand for an extended lifetime of the relays. Especially by switching direct currents a directed material transfer which is limiting the lifetime can occur. Source voltage from battery systems (e.g. automotive) can scatter in a wide range due to the actual load condition. However several work principles of the relay are functions of the source voltage. Nevertheless the application requires a steady and secure switching behavior to ensure a high life-time. The authors present a simulation model (using MATLAB/SIMULINK) describing a model switch based on a real application. It includes two different contact systems consisting two contact bolts and a contact bridge with a tilting movement. The geometrical dimensions including fabrication tolerances have been considered in the model. Three different electrical loads were investigated in combination with the model switch. A short arc model is created to simulate the current and arc voltage curves during the bouncing process. The evaluation and verification of the created short arc model is based on these three load characteristics. All simulations are validated by measurements.
    https://www.researchgate.net/publication/282795495_Simulation_of_relay_contact_bouncing_including_a_short_arc_model

Color[edit]

Calibration[edit]

Dust removal[edit]

Server[edit]

For good background, see TranslateWiki.net FAQ.

meta:Help:Magic words

  • Timestamps: 16:08, Tuesday, May 7, 2024 (UTC); 16:08; -; 20230204042400
  • Site name: Wikipedia
  • Plural: 0 counts, 1 count, 6 counts
  • Gender
    • Glrx (1 option) Only one option
    • Glrx (3 option) unknown
    • self (3 option) unknown
    • default (3 option) unknown
    • JamesBWatson (3 option) unknown
    • Liz (3 option) unknown
    • Apteva (3 option) unknown
    • Delphi234 (3 option) unknown
    • MelanieN (3 option) female

meta:Help:Parser function


Conditionals http://www.mediawiki.org/wiki/Help:Extension:ParserFunctions

  • String empty test: test string is empty
  • numeric equality test (01=1): equal
  • string equality test ("01"="1"): not equal
  • which case (casex): default result

meta:Help:calculation

  • 1+2 = 3
  • pi = 3.1415926535898
  • sin(π/6) = 0.5

Rounding

  • 22 / 7 round 2 = 3.14
  • 13 / (13 + 17 + 2) round 2 = 0.41
  • 0 = 0 : true
  • 1 = 0 : false
  • 1-1 = 0 : true
  • 22/0 : —
  • 22/7 : 3.14

Grammar[edit]

Grammar doesn't do anything on en.WP? Try on de.WP for declension.

  • he
  • she
  • it
  • one
  • May
  • May

Related plural magic word:

  • 0 letters
  • 1 letter
  • 2 letter
  • 0.3 letters

tl template tests[edit]

Documentation is wrong. See table at end of documentation.

  • {{cite}} using tl
  • {{cite}} using tl with sister=simple (sister does not work)
  • {{subst:cite}} using tls
  • {{subst:cite}} tls with sister=simple (sister does not work)
  • {{cite}} using tl2
  • {{cite}} using tl2 with sister=simple
  • {{cite}} using tlx with sister=simple (sister does not work)
  • {{cite}} using tltt with sister=simple (sister does not work)
  • {{cite}} using tlg with sister=simple (sister does not work)

Coordinates[edit]

See {{Coord}} at Helper functions. There's also Wikipedia:WikiProject Geographical coordinates#Creating new templates

Language[edit]

The current language is [undefined] Error: {{Lang}}: no text (help). (another Commons issue?)

Statement in another language: Ich bin ein Berliner. (I am a jelly donut.)

  • English: English sentence.
  • Italian: Italian sentence.
  • French: French sentence.
  • German: German sentence.
  • German: Ich bin ein Berliner.
  • (German has odd behavior of including message about German wiki. Becomes {{iw-ref}})


LangSwitch below replaced in this edit

LangSwitch (Commons?) is not connecting {{LangSwitch|en=English trouble |de=German trouble |fr=French trouble |default=Default trouble}}:English trouble

Language

  • zh 中文
  • zh-CN 中文（中国大陆）
  • zh-TW 中文（臺灣）
  • zh-HK 中文（香港）
  • zh-Hans 中文（简体）
  • zh-Hant 中文（繁體）

Research[edit]

Barcode[edit]

Pretty basic, but has other citations:

  • Method of decoding bar codes and bar code reader {{citation}}: Cite has empty unknown parameter: |inventor1-first= (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor1-last= ignored (help); Unknown parameter |patent-number= ignored (help)

Spencer[edit]

Autoload but manually cocked.

  • Spencer

Winchester[edit]

1860 Henry and Winchester '1866
  • Volcanic patent?
  • Improvement in Fire-arms {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor1-first= ignored (help); Unknown parameter |inventor1-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Drawing is "Magazine Firearm".
    Company started out as Smith and Wesson. B. Tyler Henry was an employee.
    New company appeared as Volcanic Repeating Arms Company. Volcanic purchased patent rights from Smith and Wesson.
    Financial difficulties in 1856 lead to loans from stockholders. O. F. Winchester got most of the property in the company to cover the loans.
  • Improvement in Magazine Fire-Arms {{citation}}: Unknown parameter |assign1= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Title on drawing is "Magazine Fire Arm".
    Possible safety built into lever: if trigger is back, then lever won't close
  • Improvement in Magazine Fire-Arms, March 29, 1866??? {{citation}}: Check date values in: |publication-date= (help); Unknown parameter |assign1= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Title on drawing is "Magazine Fire-Arm".
    Also Google: https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/US55012.pdf
    Reissue 9157, 13 April 1880, https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/USRE9157.pdf
    I've hit reissue several times before, but cannot find them on espacenet.
    How do I get that? {{citation}}: Empty citation (help) does not work
    Good description of 1866.
    Can use for many details such as narrowing of bolt (breech pin).
    Surprisingly, the King Gate opens out in the patent; it opens in for the 1873 (and probably 1866)
    Mortise has hole for carrier lift.
    Terminology: e.g., breech bin
    http://marauder.homestead.com/files/model66.html
    has good illustration that shows details such as carrier
    barrel says "KING'S IMPROVEMENT-PATENTED- MARCH 29 1866"
    Nelson King had succeeded B. Tyler Henry
    In 1866, some Henrys had improved gates; maybe the improved Henry was the Winchester 66
    Some reference may comment on what the catalogs named different models
    1866 Winchester had King gate and was called the "yellow boy"
  • US 57808, Winchester, O. F., "Improvement in Magazine Fire-Arms", issued 4 September 1866 
    No King gate. Credits Smith and Wesson for the tubular magazine 14 February 1854. Faults design as having open slot. This magazine design with inner tube. Also good description of the how feeding works (design details about clearance, ejection, etc) but different terminology: breech pin (bolt), piston (firing pin). Raises issue about a spring holding the lever closed as well as second spring to hold elevator up. Separate spring z for each of levers F and H (page 2, col 1).
  • 1873 (what changed?)
    Reproduction has nice pictures: http://www.realguns.com/articles/615.htm
    note carrier is flush with bottom of receiver; carrier rises above receiver at ejection.
  • 1876 (added safety, but many with safety are labeled 1873)
Winchester '73 but without safety (does that imply an early build?). Notice that the lower mortise wall is missing; that would allow an inward King gate with the gate doing all the blocking for the magazine.

I'm confused about the models. The simple explanation appears to be 1866 (rimfire) is without the safety, the 1873 has a safety, and the 1876 has a longer receiver for the longer cartridges. Winchester could not make the 1876 strong enough for the government cartridge, so it used a slightly different cartridge.

  • Sharpe, Philip B. (1995) [1938]. The Rifle in America. Odysseus Editions. pp. 222–225.

Henry rifle; rimfire .44 Henry rifle cartridge. Had a follower but no forestock. Manufactured 1860 to 1866. Apparently iron or brass receiver. Iron first, brass second.

Winchester 1866. Still rimfire? King gate. Yellow boy.

Winchester 73. Pictures vary all over the map. Don't believe it had a safety. Iron and then steel action.

.44 revised Henry rifle
.44-40 Winchester = .44 WCF = .44/40
.38-40 Winchester - necked down .44/40 = .38 WCF but not ??? (1874)
.32-20 Winchester - same length with bottleneck (1882)
.25-20 Winchester

Winchester 76. With safety. longer receiver for longer cartridges.

The government .45-70 was too long for the '73 action.
.45-75 Winchester - same as .45/70 but lighter bullet and more powder
.45-60 Winchester
.40-60 Winchester
.50-95 Express

But page 228A shows an 1876 in .44/40 (something that Sharpe does not list as a cartridge for the '76.) The receiver is longer and it has a safety.

  • US 306577, Browning, J. M. & Browning, M. S., "Magazine Fire Arm", issued 14 October 1884 
  • US 524702, Browning, J. M., "Magazine Gun", issued 21 August 1894 
  • US 549345, Browning, J. M., "Box Magazine Firearm", issued 5 November 1895  Box magazine

Maxim[edit]

  • Find Maxim patents for issues about feed
  • http://www.victorianshipmodels.com/autoMG/Maxim/secondprototype.html
  • foo, yyy {{citation}}: Check date values in: |publication-date= (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help) (guess from image on net)
  • Improvements in the Firing Mechanism of Automatic Guns {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor1-first= ignored (help); Unknown parameter |inventor1-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Improvements in Automatic Guns {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor1-first= ignored (help); Unknown parameter |inventor1-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help), gas action for machine guns
  • Improvements in Automatic Machine Guns {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor1-first= ignored (help); Unknown parameter |inventor1-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help), breech mechanism of machine gun
The Book of the Machine Gun By Frederick Victor Longstaff, A. Hilliard Atteridge
footnote, page 24
The first of the Maxim patents is dated July 16 1883 No 3493 It is for an invention of improvements in machine or battery guns and in cartridges for the same and other firearms In the No 3493 the feed block was in the bottom of body There are further patents dated 1884 January 3 No 606 and February 23 No 3844 for machine guns There is a patent May 23 of the same year No 9407 for a cartridge feed and October 2 No 13113 for the lubrication of machine guns There are further patents dated January 29 1885 No 1307 and July 8 1885 No 8281 this last being the first patent in which the machine gun assumed much the present Maxim form There are further patents in 1890 1892 1894 and 1895 in the names of Sir HS Maxim or jointly with his assistant L Silverman The 1895 patent No 5864 dated March 20 describes the form of lock now in use.

Poulsen arc[edit]

Describe the gain mechanism.

GoogleWiFi[edit]

Junk[edit]

Gaseous conductors[edit]

Quick coupler[edit]

Printer technology[edit]

Grit versus knurled wheel printer controversy. "B & L agrees that the use of grit provides great advantages over a knurled wheel. In particular, grit is much harder and sharper than the points of a knurled wheel, and so is especially effective in creating indentations in the paper." (909 F.2d 1464 @ ¶ 16.)

Apparently, the varying compression of the rubber rolls hampered the plotter's ability to achieve accurate registration. Eventually, the idea of using rubber wheels was discarded as imperfect. Instead, LaBarre considered knurling metal rolls, spiraling sandpaper on the outside of them with the grit out, or impregnating epoxy coated wheels with grit particles. (@597)
Lloyd first learned of the LaBarre invention when one of his co-workers at Versatec returned from a trade show and described the drive system invented by LaBarre to a group of fellow engineers. "[I]t was our impression that someone had finally solved the problem of how to move paper accurately." (@603–604)
Mr. Yeiser himself, who would qualify as one skilled in the art, could not build a successful X-Y plotter. (@606)
In 40 passes across the paper, the cumulative error was less than .003 inches. Inspired, he wrote in his diary, "They work `Eureka'." (@597)
knurling is diamond pattern. "The drive wheels for moving paper are made of rubber or metal; and can be smooth or knurled." (quoting patent) 722 F.Supp @600
  • U.S. Pat. No. 4,384,298 (LaBarre) (Patent examiner knew of Yeiser, so presumption of validity is strong. ¶ 14)

LaBarre was an "apparatus" patent. Differences in performance are not relevant. (¶ 18):

Claim 1 of LaBarre is an apparatus claim, and apparatus claims cover what a device is, not what a device does. An invention need not operate differently than the prior art to be patentable, but need only be different. ... By focussing its argument on the alleged lack of "operational differences" between the device of the claim language and the prior art, B & L has utterly failed to point out any evidence, much less clear and convincing evidence, of why one skilled in the art would replace the knurled wheel of Yeiser with one having a random pattern, size and shape of rough spots.

Toroid winding machine[edit]

Interesting machine that winds toroid cores. Basic plan is to get a loop of wire rotating through toroid; coil falls off loop and gets wrapped around the toroid; toroid spins to evenly distribute the wire.

Baker (1959) uses a soldering iron (for starting?). I don't see slip off is regulated. Loop is distored to travel through toroid.

Oshima (1964) has some slip cups above and below the toroid (10, 10'). Slip may be through bobbin.

Fahrbach (1967)

Potthoff (1978) Hand crank, drive belt pulled inside so loop can travel through torid.

  • US 841305 Atwood Jan 1907
  • US 3000580 Matovich, Jr Sept 1961
  • US 3451631 Tillman June 1969
  • US 3811629 Sedgewick May 1974
  • US 545674 Burns 3 Sept 1895
  • US 2102692 Franz 21 Dec 1937
  • Ring Winding Machine {{citation}}: Unknown parameter |air-date= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Winding Machine for Magnetic Cores of Small Size {{citation}}: Unknown parameter |air-date= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Method and Apparatus for Winding Toroidal Coils {{citation}}: Unknown parameter |air-date= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Toroidal Core Winder {{citation}}: Unknown parameter |air-date= ignored (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)

Patents[edit]

Tektronix[edit]

  • Electrical Probe {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Transmission Line Termination Circuit {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Shielded cable having auxiliary signal conductors formed integral with shield {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)

Noise[edit]

  • US 6268735, x, "Noise source module for microwave test systems", issued July 31, 2001 
  • US 3619780, x, "Transistor Noise Measuring Apparatus", issued November 9, 1971 
  • US 5191294, x, "Measuring noise figure and y-factor" . Wiltron
  • US 3302116, x, "Signal plus noise to noise measuring equipment" 
  • US 4808912, x, "Six-port reflectometer test arrangement and method including calibration" . Marconi
  • US 5434511, x, "Electronic microwave calibration device" . ATN Microwave, Inc.
  • US 5416422, x, "Apparatus and method for determining single sideband noise figure from double sideband measurements" . Hewlett-Packard
  • US 4905308, x, "Noise parameter determination method" 
  • US 2901696, x, "Arrangement for automatic and continuous measuring of the noise factor of an electric device", issued August 25, 1959  Gated.
  • US 2620438, x, "Noise-factor meter", issued December 2, 1952  Gated.
  • US 2691098, x, "Automatic noise figure meter", issued October 5, 1954 
  • US 6066953, x, "Architecture for RF signal automatic test equipment" . Teradyne. Network Analyzer?
  • US 4905308, x, "Noise parameter determination method" 
  • US 5434511, x, "Electronic microwave calibration device" 
  • US 4858160, x, "System for setting reference reactance for vector corrected measurements" 
  • US 3209279, mumble, "mumble" 

Directional bridge[edit]

  • Frequency selective improvement of the directivity of a return loss bridge {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Hand draw figures. Improve directivity. Tektronix.
  • Bridge for Measuring the Reflection Coefficient {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Rhode-Schwarz.
  • Directional Sampling Bridge {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Stanford University. 100GHz.
  • Three section termination for an R.F. triaxial directional bridge {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Hewlett-Packard. Earlier of 4720677.
  • R. F. Triaxial Directional Bridge {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Hewlett-Packard
  • Vector network analyzer with integral processor {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). VNA. Hewlett-Packard. Also use FFT to get time domain. 8510?
  • Spectrum analyzer having enterable offsets and automatic display zoom {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-first= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Spectrum analyzer. Older VNA? Hewlett-Packard. Was cited in US 4816767.
  • Return Loss Test Set {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). doubled bridge? GTE Automatic Electric Lab.
  • Dual directional bridge and balun used as reflectometer test set {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Test set. 300kHz to 3.0GHz. Hewlett-Packard.

Directional coupler[edit]

Spot welding[edit]

It's a small world. I was looking at spot welding, and someone provided a early Popular Mechanics article that made me wonder who invented spot welding. A long time later, I'm looking at oscillators and come across Elihu Thomson with a singing arc -- something I'd heard about in my youth. Then I find out he invented resistance welding:

  • Apparatus for Electric Welding, 29 March 1886 {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventorlink= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
  • Method of Electric Welding, 14 June 1890 {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventorlink= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help)
Thomson proposed transformers, batteries, and dynamos for the high current source in the first patent.


  • Welding by the Charge of a Condenser {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). A capacitor (aka condenser) is used to limit the energy delivered during a spot welding operation. Capacitor is charged rather than discharged. Goal is delicate work.
  • Dual Current Condenser Welder {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Second condenser is discharged into the weld. "I have discovered that the weld is considerably improved if a second condenser is provided and connected in such a manner that it discharges into the weld...."
  • Welding System {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Capacitive discharge with mechanical energy storage and step down transformer. Goal is high peak powers.
  • Long-Lived Impulse Transformer {{citation}}: Cite has empty unknown parameter: |inventor-first= (help); Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Construction details for welding transformer.
  • Portable Spot Welding Tool {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Capacitive discharge spot welder with stepdown transformer. Goal is delicate work.
  • Spot Welder {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help).
  • High Current, Low Impedance Resistance Welding Device {{citation}}: Unknown parameter |country-code= ignored (help); Unknown parameter |inventor-first= ignored (help); Unknown parameter |inventor-last= ignored (help); Unknown parameter |inventor2-last= ignored (help); Unknown parameter |issue-date= ignored (help); Unknown parameter |patent-number= ignored (help). Goal is welding seams that have a resistance in the 1 to 100 micro-ohm range.

Digital caliper[edit]

hyphen[edit]

For just pages=, a hyphen is changed to an endash. U+2011 or #8209; is a nonbreaking hyphen. (- ‑ − – —)

  • John Q. Public. Some Book. 13-14. (at=)
  • John Q. Public. Some Book. p. 13-14. (page=)
  • John Q. Public. Some Book. pp. 13–14. (pages=)
  • John Q. Public. Some Book. pp. 13-14–13-20. (pages=) (wrong)
  • John Q. Public. Some Book. pp. 13&#8209, 14–13&#8209, 20. (pages= using non-breaking hyphen; position shifts)

Descreening[edit]

Scanned halftone image

Scanned halftone images have problems. Moiré patterns.

There was a paper by Picture Elements, Inc. about filterning the images, but I cannot find it. DFTs found the screen frequency and screen angle.

Another PEI document

Advice on web has stuff about Gaussian blur or other image processing filters in programs such as Photoshop.

  • Photoshop:
  • CorelPAINT: effects/noise/remove Moiré.

Validation[edit]

It is desirable to have all XML files validate. It was even a goal of writing DTDs. Some DOM parsers will raise an exception if the DTD is violated. (They also take more compute time to do the validation. I had to find out how to turn off Java's XML validation because it was taking 6 seconds to validate a file.) Even if validation is not done all the time, it is reassuring to know that an XML file validates.

SVG files should display {{Valid SVG}}, but that is where some trouble arises. Unfortunately, including information (such as Common Core licensing) in a metadata element often prevents validation because the DTD is strict. Furthermore, XML extensions are often sprinkled throughout an XML file. Translation markup often want to identify elements that should not be translated with the attribute its:translate="no". That attribute is not in the underyling DTD, so it raises an error. One way around the problem is to extend the underlying DTD; HTML 5, for example, now has a translate attribute. That's not universal; SVG 2.0 does not have a translate attribute. Many XML-based file formats now recommend against using a DOCTYPE declaration. Apparently, a DTD was never written for SVG 1.2. DTD were also developed before XML namespaces. Namespaces give name isolation, but they confound the DTD issue.

Look at schemas. Relax.

IIRC, there are some namespace twists. Elements without a namespace prefix belong to the default namespace (whatever xmlns=... dictates), but attributes without a namespace prefix belong to the null namespace.

Many experiments provide insight.

W3 validator[edit]

Many XML validators are available. DOM processors have internal validators. There are also web-based validators. W3 has one at

{{Valid SVG}} uses the W3 validator, so it deserves some attention.

which provides the following example (validates as XML due to SYSTEM; added encoding="UTF-8"): 

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE rdf:RDF SYSTEM "http://dublincore.org/2000/12/01-dcmes-xml-dtd.dtd">
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
         xmlns:dc ="http://purl.org/dc/elements/1.1/">
 <rdf:Description rdf:about="http://dublincore.org/">
   <dc:title>Dublin Core Metadata Initiative - Home Page</dc:title>
   <dc:description>The Dublin Core Metadata Initiative Web site.</dc:description>
   <dc:date>1998-10-10</dc:date>
   <dc:format>text/html</dc:format>
   <dc:language>en</dc:language>
   <dc:contributor>The Dublin Core Metadata Initiative</dc:contributor>
   <dc:contributor xml:lang="fr">L'Initiative de métadonnées du Dublin Core</dc:contributor>
   <dc:contributor xml:lang="de">der Dublin-Core Metadata-Diskussionen</dc:contributor>
 </rdf:Description>
</rdf:RDF>

DOCTYPE implies XML validation[edit]

Metadata version that fails SVG 1.1 validation (validates as XML if DOCTYPE removed; does not complain about <gubbish/> tag, so no SVG validation done!): 

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd" >

<svg version="1.1" baseProfile="full"
     xmlns="http://www.w3.org/2000/svg"
     xmlns:xlink="http://www.w3.org/1999/xlink"
     xmlns:ev="http://www.w3.org/2001/xml-events"
     xmlns:its="http://www.w3.org/2005/11/its"
     width="1200"
     height="500"
     viewBox="0 0 1200 500"
     overflow="visible"
     enable-background="new 0 0 1200 500"
     xml:space="preserve">

<title>Energy to ionize the first electron</title>
<desc>Energy required to ionize the first electron from an atom.</desc>
<metadata id="license"
          xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
          xmlns:dc="http://purl.org/dc/elements/1.1/"
          xmlns:cc="http://creativecommons.org/ns#">
  <rdf:RDF>
    <cc:Work rdf:about="http://commons.wikimedia.org/wiki/File:First_Ionization_Energy.svg">
      <dc:format>image/svg+xml</dc:format>
      <dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
      <cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/" />
      </cc:Work>
    <cc:License rdf:about="http://creativecommons.org/licenses/by-sa/3.0/">
      <cc:permits rdf:resource="http://creativecommons.org/ns#Reproduction" />
      <cc:permits rdf:resource="http://creativecommons.org/ns#Distribution" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#Notice" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#Attribution" />
      <cc:permits rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#ShareAlike" />
      </cc:License>
    </rdf:RDF>
  </metadata>
</svg>

Another validator

Content-Type interaction[edit]

Sometimes the validator decides the SVG is SVG 1.1+XHTML+MathML 3.0 and ignores Inkscape and RDF, but it didn't for me in the sections below. I copied the header and tried the stuff immediately below, but it was just checked for XML.

I took a file that validates as SVG 1.1+XHTML+MathML 3.0, copied its contents into the clipboard, and then did a direct input validation. It validated as XML.

I suspect there is no DTD for SVG 1.1+XHTML+MathML 3.0. Consequently, there is no DOCTYPE that can be entered to specify that checking. Instead, the validator probably cues off the webserver's Content-Type. With direct input, the validator may assume text/xml rather than something like image/svg+xml. Checked that WP returns image/svg+xml, so I believe this hypothesis. 

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- Created with Inkscape (http://www.inkscape.org/) -->
<svg xmlns:dc="http://purl.org/dc/elements/1.1/"
     xmlns:cc="http://creativecommons.org/ns#"
     xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
     xmlns:svg="http://www.w3.org/2000/svg"
     xmlns="http://www.w3.org/2000/svg"
     xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
     xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
     version="1.1" width="958.69" height="592.78998" id="svg2275"
     inkscape:version="0.48.2 r9819"
     sodipodi:docname="Map_of_USA_with_state_names-en.svg">
<metadata id="metadata362">
  <rdf:RDF>
    <cc:Work rdf:about="">
      <dc:format>image/svg+xml</dc:format>
      <dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
    </cc:Work>
  </rdf:RDF>
</metadata>
</svg>

SVG[edit]

The code from SVG 1.1 spec (http://www.w3.org/TR/SVG/metadata.html) validates as XML (no SVG DOCTYPE). 

<?xml version="1.0" standalone="yes"?>
<svg width="4in" height="3in" version="1.1"
    xmlns = 'http://www.w3.org/2000/svg'>
    <desc xmlns:myfoo="http://example.org/myfoo">
      <myfoo:title>This is a financial report</myfoo:title>
      <myfoo:descr>The global description uses markup from the
        <myfoo:emph>myfoo</myfoo:emph> namespace.</myfoo:descr>
      <myfoo:scene><myfoo:what>widget $growth</myfoo:what>
      <myfoo:contains>$three $graph-bar</myfoo:contains>
        <myfoo:when>1998 $through 2000</myfoo:when> </myfoo:scene>
   </desc>
    <metadata>
      <rdf:RDF
           xmlns:rdf = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
           xmlns:rdfs = "http://www.w3.org/2000/01/rdf-schema#"
           xmlns:dc = "http://purl.org/dc/elements/1.1/" >
        <rdf:Description rdf:about="http://example.org/myfoo"
             dc:title="MyFoo Financial Report"
             dc:description="$three $bar $thousands $dollars $from 1998 $through 2000"
             dc:publisher="Example Organization"
             dc:date="2000-04-11"
             dc:format="image/svg+xml"
             dc:language="en" >
          <dc:creator>
            <rdf:Bag>
              <rdf:li>Irving Bird</rdf:li>
              <rdf:li>Mary Lambert</rdf:li>
            </rdf:Bag>
          </dc:creator>
        </rdf:Description>
      </rdf:RDF>
    </metadata>
</svg>

Extending the DTD[edit]

Play some games...

Probably should not say standalone="yes". Problems with defaulted from external subset.

Gets one error -- does this mean metadata element is not ANY? The metadata rule is (#PCDATA ...) 

Line 25, Column 58: document type does not allow element "rdf:RDF" here 
           xmlns:dc = "http://purl.org/dc/elements/1.1/" >✉ 
The element named above was found in a context where it is not allowed. This could mean that you have incorrectly nested elements -- such as a "style" element in the "body" section instead of inside "head" -- or two elements that overlap (which is not allowed). 

One common cause for this error is the use of XHTML syntax in HTML documents. Due to HTML's rules of implicitly closed elements, this error can create cascading effects. For instance, using XHTML's "self-closing" tags for "meta" and "link" in the "head" section of a HTML document may cause the parser to infer the end of the "head" section and the beginning of the "body" section (where "link" and "meta" are not allowed; hence the reported error). 

<?xml version="1.0" ?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[
<!ELEMENT rdf:RDF ANY>
<!ELEMENT rdf:Description ANY>
<!ATTLIST rdf:Description dc:title       CDATA #IMPLIED
                          dc:description CDATA #IMPLIED
                          dc:publisher   CDATA #IMPLIED
                          about          CDATA #IMPLIED
                          dc:format      CDATA #IMPLIED
                          dc:language    CDATA #IMPLIED
                          dc:date        CDATA #IMPLIED>
<!ELEMENT rdf:Bag ANY>
<!ELEMENT rdf:li ANY>

<!ELEMENT dc:creator ANY>

<!-- and the trick... first definition wins -->
<!ENTITY % SVG.metadata.extra.content "| rdf:RDF" >

]>
<svg width="4in" height="3in" version="1.1"
    xmlns = 'http://www.w3.org/2000/svg'>
    <desc>Just some simple text</desc>
    <metadata>
      <rdf:RDF
           xmlns:rdf = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
           xmlns:rdfs = "http://www.w3.org/2000/01/rdf-schema#"
           xmlns:dc = "http://purl.org/dc/elements/1.1/" >
        <rdf:Description rdf:about="http://example.org/myfoo"
             dc:title="MyFoo Financial Report"
             dc:description="$three $bar $thousands $dollars $from 1998 $through 2000"
             dc:publisher="Example Organization"
             dc:date="2000-04-11"
             dc:format="image/svg+xml"
             dc:language="en" >
          <dc:creator>
            <rdf:Bag>
              <rdf:li>Irving Bird</rdf:li>
              <rdf:li>Mary Lambert</rdf:li>
            </rdf:Bag>
          </dc:creator>
        </rdf:Description>
      </rdf:RDF>
    </metadata>
</svg>

For SVG data-* attributes, extend the SVG 1.1 DTD attributes with "SVG.External.extra.attrib". 

<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[
<!ENTITY % SVG.External.extra.attrib "
  data-area CDATA #IMPLIED
  data-src  CDATA #IMPLIED
  ">
 ]>

Unicode normalization[edit]

There can be problems within a file that make it fail validation. For example, problems include invalid NCNAMEs in identifier attributes. An interesting problem was unnormalized Unicode strings raising errors for an SVG file. There are methods than can used to normalize Unicode strings.

See extensive discussion on Commons:User_talk:Delphi234.

That file passes SVG validation after several fixes.

  • SVG version is 1.0 rather than 1.1 (validator only wants 1.1) (warning only; not fixed)
  • There are some spurious text content in the switch elements (i.e., bare characters rather than within text elements. (deleted extra characters)
  • Some warnings: text strings are not in Unicode Normalization Form C. (see http://www.unicode.org/reports/tr15/) These are for Indic Kannada codepoint sequences. See Kannada (Unicode block). (fixed; faked out somehow; cut and paste from SVG file shows not NFC.)
    systemLanguage="kn" ಮೊಂಟಾನಾ (Montana) (error combining vowel AA at end 0CBE)
    systemLanguage="kn" ಕೊಲೊರಾಡೋ (Colorado) (error combining length mark at end 0CD5)
    systemLanguage="kn" ದಕ್ಷಿಣ ಡಕೋಟಾ (South Dakota... but ಉತ್ತರ ಡಕೋಟ North Dakota was OK? ) (error combining vowel AA at end 0CBE)
    Google translate(Montana ) = ಮೊಂಟಾನಾ but alternative ಮೊಂಟಾನ
    Google translate(Colorado ) = ಕೊಲೊರಾಡೋ but alternative ಕೊಲರಾಡೋ
    Google translate(South Dakota) = ದಕ್ಷಿಣ ಡಕೋಟಾ (which is on used in SVG) but alternative ದಕ್ಷಿಣ ಡಕೋಟ
    Google translate(North Dakota) = ಉತ್ತರ ಡಕೋಟ but alternative ಉತ್ತರ ಡಕೋಟಾ
    Apparently SVG check wants all text in NFC. A simple fix would be a canonical decomp followed by canonical composition. Look at code to do that:
    Library https://github.com/walling/unorm
    ECMA script (later versions) provide those tools, and walling/unorm has an emulation of ECMA routine
    https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/String/normalize
    str.normalize("NFC" or "NFD" or "NFKC" or "NFKD")
    I don't see the problem. ECMAScript trials do not give me a clue about what is going wrong because normalize gives same result as the offending string, which implies the offending string is in NFC. A possible clue is I believe North Dakota would be in the SVG file with "kn" translation, and there is no complaint about that script. I used Google translate to translate both of them, and found they differ in their composition characters. In Kannada, consonant characters have a default vowel; the consonant can be modified to delete or modify the vowel.
    Maybe the SVG checker is using the fast NFC validation test and getting fooled?
    Montana has two composition characters, but the Unicode table does not have a single composition character for that combination.
    U+0CCA Vowel sign O
    U+0C82 Ansuvara (apparently no combined forms)
    ECMA NFC is idempotent, but SVG check fails.
    Colorado.
    This could be separate length marker as in Dakota.
    U+0CCB Vowel sign OO
    Dakota. The Unicode tables seems to say that composition for Dakota should be complete; South Dakota has a single composition; North Dakota has two composition characters.
    U+0CCB Vowel sign OO (used in South Dakota)
    U+0CCA U+0CD5 Length mark (used in North Dakota)
    ECMA NFC is idempotent on South Dakota, but SVG check fails.
    ECMA NFC reduces two composition characters to one on North Dakota (length marker separate), but SVG check presumably succeeds.
  • I notice there is no title element
  • I notice broken style font-family: (spaces must be quoted) (fixed the one below; others exist) 
    <g xml:space="preserve" style="font-size:13.08702374px;font-style:normal;font-variant:normal;font-weight:bold;font-stretch:normal;text-align:start;line-height:100%;writing-mode:lr-tb;text-anchor:middle;fill:#000000;fill-opacity:0.75;stroke-width:3pt;font-family: Nimbus Sans L,'Arial Narrow',sans-serif;Sans L',sans-serif;-inkscape-font-specification:Nimbus Sans L Bold">

Font shifts needed? bn versus bpy

  • ওয়াশিংটন
  • ৱাশিংটন

License[edit]

There is a large issue about licensing. I do not understand why wiki files do not embed the license terms and a link to Commons in its exported files that support such information. TIFF, JPEG/Exif, and SVG/XML can carry metadata and linking.

Sadly, I've seen somebody take a Commons file, edit it, and upload it with a claim that it is their own. The metadata should be scavenged from the source file (and perhaps not be alterable). If there is no metadata, then it should be inserted. If a file is copied from Commons, it should be able to phone home.

CC watermarks appear discouraged.

Some activity on commons: Commons:Commons:Structured data. de.WP has funding....

There's a category issue. Maybe get categories from original and then filter?

Ah, there is a difference between uploading a new file and updating an existing file.

On meta:

WMF wants free images to stay free, but that seems in conflict with typical practice.

Wikidata copyright[edit]

What was the name of the Commons proposal? There was a recent RfC with de.WP having source of funding.

See

Book[edit]

Let Qxxxx be a copyrightable work.

  1. Get its publication date.
  2. Get its publication country.
  3. Get the author property.
  4. Get author's date of death.
  5. Deduce copyright status.

Botanical illustrations in books[edit]

Many illustrations have been credited to a book's author rather than the illustrator.

In addition, many illustrations are missing {{PD-scan}}.

Walther Otto Müller (Walther Otto Müller (Q21521878))

Otto Wilhelm Thomé Otto Wilhelm Thomé (Q76714)

  • Wikidata claims he was an illustrator, but I don't think this is true.

No Q for Franz Köhler the botanist.

  • Köhler probably not an illustrator

Painting[edit]

The Musée d'Orsay:

It describes the painting

So I get the creation date and the date of death. 120 years after death is 1997. France sweat-of-the-brow copyright cases have gone both ways.

Metadata[edit]

Metadata is a broad topic, but the basic notion is putting information, such as exposure, location, and copyright information, into the file that contains the primary data. Most picture formats have tags for metadata.

The primary focus is putting copyright information into an rdf:RDF element inside of SVG's metadata element in such a way that the SVG file will still validate.

A secondary focus would be adding ITS information, but that will probably cause validation to fail.

Dublin core[edit]

Dublin core defines the vocabulary for representing relationships such as author.

Creative Commons[edit]

Primary license expression on Wikipedia follows Creative Commons and is usually expressed as a Creative Commons license. The license variants are expressed by some (possibly exclusive) properties:

  • CC Creative Commons
  • CC0 public domain (no rights reserved)
    • -BY Attribution required
    • -NC Non commercial
    • -SA Share Alike
    • -ND No derivatives

There are different versions of these licenses (including ported versions which are disappearing). URIs exist for different licenses:

These URIs link to HTML descriptions of the licenses. I expect to get an RDF description, but instead there is a link that I'd expect to be to a straight RDF description. However, I cannot find the linked file.

How these licenses are claimed in different files is covered in ccREL: The Creative Commons Rights Expression Language :

It covers several formats such as HTML, RDF, and SVG.

  • In HTML, a text statement with a hyperlink
<div about="http://lessig.org/blog/" xmlns:cc="http://creativecommons.org/ns#">
This page, by
<a property="cc:attributionName" rel="cc:attributionURL" href="http://lessig.org/">Lawrence Lessig</a>,
is licensed under a
<a rel="license" href="http://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution License</a>.
</div>

ccREL describes the rights as a work and a license.

Separate RDF files.

The representation is now Work + possibly a license.

Work triples/properties

  • Work xhtml:license URI (same as cc:license)
  • Work dc:title ...
  • Work cc:atttributionName ...
  • Work cc:attributionURL ...
  • Work dc:type (http://purl.org/dc/dcmitype/Text StillImage Image ...)
  • Work dc:source
  • Work cc:morePermissions

License triples/properties

  • l cc:permits (cc:Reproduction, cc:Distribution, cc:DerivativeWorks)
  • l cc:prohibits (cc:CommercialUse)
  • l cc:requires (cc:Notice, cc:Attribution, cc:ShareAlike, cc:SourceCode)
  • l cc:jurisdiction
  • l cc:deprecatedOn
  • l cc:legalCode

Specify value with rdf:resource attribute or element contents. 

  <metadata>
    <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
             xmlns:dc="http://purl.org/dc/elements/1.1/"
             xmlns:cc="http://creativecommons.org/ns#">
      <cc:Work rdf:about="">
        <dc:title>Diagram Verfassung Bosch</dc:title>
        <dc:source rdf:resource="https://commons.wikimedia.org/wiki/File:Diagram_Verfassung_Bosch.png"/>
        <cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Habakuk"/>
        <cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Lasse_Havelund"/>
        <cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Perhelion"/>
        <cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Glrx"/>
        <cc:attributionURL rdf:resource="https://commons.wikimedia.org/w/index.php?title=File%3ABosch_Composition.svg"/>
        <dc:format>image/svg+xml</dc:format>
        <dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
        <cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/deed.en" />
      </cc:Work>
    </rdf:RDF>
  </metadata>

RDFa.

  • about="" means the current document.
  • property="cc:foo" sets the predicate to be cc:foo and the object is the text content of the element
  • rel="cc:foo" sets the predicate and the object is the href
  • rel="license" sets the predicate and the object is the href
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd" >

<svg version="1.1" baseProfile="full"
     xmlns="http://www.w3.org/2000/svg"
     xmlns:xlink="http://www.w3.org/1999/xlink"
     xmlns:ev="http://www.w3.org/2001/xml-events"
     xmlns:its="http://www.w3.org/2005/11/its"
     width="1200"
     height="500"
     viewBox="0 0 1200 500"
     overflow="visible"
     enable-background="new 0 0 1200 500"
     xml:space="preserve">

<title>Energy to ionize the first electron</title>
<desc>Energy required to ionize the first electron from an atom.</desc>
<metadata id="license"
          xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
          xmlns:dc="http://purl.org/dc/elements/1.1/"
          xmlns:cc="http://creativecommons.org/ns#">
  <rdf:RDF>
    <cc:Work rdf:about="http://commons.wikimedia.org/wiki/File:First_Ionization_Energy.svg">
      <dc:format>image/svg+xml</dc:format>
      <dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
      <cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/" />
      </cc:Work>
    <cc:License rdf:about="http://creativecommons.org/licenses/by-sa/3.0/">
      <cc:permits rdf:resource="http://creativecommons.org/ns#Reproduction" />
      <cc:permits rdf:resource="http://creativecommons.org/ns#Distribution" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#Notice" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#Attribution" />
      <cc:permits rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
      <cc:requires rdf:resource="http://creativecommons.org/ns#ShareAlike" />
      </cc:License>
    </rdf:RDF>
  </metadata>
</svg>


Embedded RDF. There's a standard for XSL transformation to extract the embedded RDF. 

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- Created with Inkscape (http://www.inkscape.org/) -->
<svg xmlns:dc="http://purl.org/dc/elements/1.1/"
     xmlns:cc="http://creativecommons.org/ns#"
     xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
     xmlns:svg="http://www.w3.org/2000/svg"
     xmlns="http://www.w3.org/2000/svg"
     xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
     xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
     version="1.1" width="958.69" height="592.78998" id="svg2275"
     inkscape:version="0.48.2 r9819"
     sodipodi:docname="Map_of_USA_with_state_names-en.svg">
<metadata id="metadata362">
  <rdf:RDF>
    <cc:Work rdf:about="">
      <dc:format>image/svg+xml</dc:format>
      <dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
    </cc:Work>
  </rdf:RDF>
</metadata>
</svg>

RDF editor for SVG[edit]

There's got to be an Resource Description Framework (RDF/XML) editor around somewhere.

There's a file on Commons, and somebody copies it and uses it. There's no guarantee of attribution. There could be RDF info, and some of that could be automatically inserted. Something basic would just make a URL point back to Commons. Further efforts would make sure the file and the Commons description refer to the same sharing license.

Consider the tag set.

Also, JPEGs should be checked their Exchangeable image file format copyright information, but I'm confused about this. Copyright appears to be in TIFF and WAV files, but I'm not clear about JPEG. It may be that TIFF style information is placed in the JPEG#Syntax and structure file as APP1.

RDF DTD[edit]

Several comments on the web about no RDF DTD. Here's an old DTD attempt:

which, if examined, says the following: 

<!-- 
     This DTD fragment is identified by and should be available at
         http://www.w3.org/Metadata/RDF/Group/9710/rdf.dtd

     Copyright 1997, 1999 by MIT, INRIA, and Keio
     $Id: rdf.dtd,v 1.3 1999/02/08 17:41:51 connolly Exp $

     It is a transcription of the grammar in
       Resource Description Framework (RDF) Model and Syntax
       Ora Lassila and Ralph R. Swick
       Version 0.2, 01-Oct-1997
       http://www.w3.org/Metadata/RDF/Group/9710/WD-rdf-syntax-971002
       updated for
       http://www.w3.org/TR/1999/PR-rdf-syntax-19990105


     The grammar is presented in a different order because SGML
     requires parameter entities to be declared before they are referenced.

     Typical usage:

     <!DOCTYPE rdf:RDF system
             "http://www.w3.org/XML/9710rdf-dtd/rdf.dtd">
     <rdf:RDF>...</rdf:RDF>

     Documents that include this DTD fragment by reference
     SHOULD NOT reference parameter entities declared herein
     (e.g. %RDF;).

     These names are intended to be "private."
     Only the element and attribute names are intended to be exported.

     The parameter entity rdf-namespace-prefix is intended to
     be "imported" ala:

    <!DOCTYPE [
    <!ENTITY % rdf-namespace-prefix "blort:">
    <!ENTITY % rdf-dtd system
             "http://www.w3.org/XML/9710rdf-dtd/rdf.dtd">
    %rdf-dtd;
    ]>
    <blort:RDF>...</blort:RDF>

     Oh for real SGML/XML namespaces... ;-)
-->

References

GRDDL[edit]

GRDDL ("Griddle") is used to extract RDF from XML files. Gleaning Resource Descriptions from Dialects of Languages.

Commons Structured Data[edit]

RDF and Maps[edit]

Port Royal earthquake in 1692.
Port Royal ca 1785.
Significantly nonlinear coordinate mapping. Lambert Conformal Conic Projection.
Austria Hungary 1911 map with grid lines but no specified mapping.

RDF can be used in maps to identify coordinate transformations. RDF can also be used to designate a past period.

Take, for example, an old map of Port Royal.

Documents

The Wikimaps warper is not using RDF. After warping, the original image on Commons is unchanged. There is a link to the warper page that has a mapping table.

Want a query that will grab nearby cities. Filter on longitude and latitude? long 8 to 28; lat 41 to 52 

SELECT DISTINCT ?city ?cityLabel (SAMPLE(?location) AS ?location) (MAX(?population) AS ?population)
WHERE
{
  ?city wdt:P31/wdt:P279* wd:Q515;
        wdt:P625 ?location;
        wdt:P1082 ?population.
  FILTER(?population >= 500000).

  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
GROUP BY ?city ?cityLabel

SELECT ?place ?placeLabel ?location
WHERE
{
  # Budapest coordinates
  wd:Q1781 wdt:P625 ?Loc .
  SERVICE wikibase:around {
    ?place wdt:P625 ?location .
    bd:serviceParam wikibase:center ?Loc .
    bd:serviceParam wikibase:radius "300" .
  } .
  # Is a city - now human settlement
  FILTER EXISTS { ?place wdt:P31/wdt:P279* wd:wd:Q486972 } .
  ?place wdt:P1082 ?population .
  FILTER (?population > 50000) .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}

Translation[edit]

World is not simple[edit]

I thought that Unicode would simplify the translation problem: one could use a single font for all the translations. Turns out, different languages/locales prefer different character shapes. That means that some translations should change fonts or do character selection within a font.

CSS and OpenType[edit]

See CSS Fonts Module Level 4.0. Uses OpenType language names: https://docs.microsoft.com/en-us/typography/opentype/spec/languagetags

Allegedly Firefox only, but cannot get it to work. Usually, the content language specification is enough. Sometimes font-language-override is used to render an unsupported content language with a similar language.

Same characters (I do not see a difference)

  • Russian alphabet
    Russian: АаБбВвГгДдЕеЁёЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЫыЬьЭэЮюЯя
  • Bulgarian alphabet
    Bulgarian АаБбВвГгДдЕеЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЬьЮюЯя.
    АаБбВвГгДдЕеЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЬьЮюЯя
  • Side by side 
    АаБбВвГгДдЕеЁёЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЫыЬьЭэЮюЯя
    АаБбВвГгДдЕе  ЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъ  Ьь  ЮюЯя

Same word in Spanish Señora and Italian Señora.

Try historical forms (looking for Jeſuit and Congreſs):

  • Jesuit Congress ſ (default)
  • Jesuit Congress ſ (Bodoni)
  • Jesuit Congress ſ (Bodoni MT)
  • Jesuit Congress ſ (Caslon)
  • Jesuit Congress ſ (Garamond)
  • Jesuit Congress ſ (Times New Roman)
  • -moz-
  • Jesuit Congress ſ (default)
  • Jesuit Congress ſ (Bodoni)
  • Jesuit Congress ſ (Bodoni MT)
  • Jesuit Congress ſ (Caslon)
  • Jesuit Congress ſ (Garamond)
  • Jesuit Congress ſ (Times New Roman)

See Long s.

Unicode: LATIN SMALL LETTER LONG S &#x17F: "ſ", so Jeſuit and Congreſs.

Such a change could be done with CSS, but that gets us into different problems. The CSS :lang selector (e.g., p:lang(zh) {..}). The xml:lang and HTML lang attributes should track. Should make some test code to see if CSS looks at both attributes. HTML lang attributes are single-valued. HTML 5 seems to want xml:lang to disappear in favor of lang. SVG is more complicated (and CSS rule may not be available). The xml:lang attribute is single valued. The SVG lang attribute appears in SVG 2.0 and is single valued. The systemLanguage attribute is multi-valued — a reason to shatter the attribute. SVG switch uses the systemLanguage attribute. First consider single-valued attribute solutions. That could use attribute value selection to set the font. Alternatively, CSS to install xml:lang attribute depending on systemLanguage attribute (check). Alternatively, one could always set a language attribute on nodes that need font selection. (There should be an ITS rule that avoids translating children of a switch.)

Oriental Languages[edit]
  • Another diversion. I can look in the font directory (C:/Windows/Fonts) and find fonts for specific scripts. Characteristics given at bottom of page.
    Deng Xian Regular (Han Simplified)
    Meiryo (Japanese)
    Meiryo UI (Japanese)
    Microsoft JhengHei (Han Traditional)
    Microsoft MHei (not stated)
    Microsoft NeoGothic (Korean?)
    Microsoft YaHei (Han Simplified)
    MingLiU_HKSCS-ExtB Regular (Han Traditional)
    MingLiU-ExtB Regular (Han Traditional)
    MS Gothic Regular (Japanese)
    MS Mincho Regular (Japanese)
    MS PGothic Regular (Japanese)
    MS PMincho Regular (Japanese)
    NSimSun Regular (Han Simplified)
    PMingLiU-ExtB Regular (Han Traditional)
    SimSun Regular (Han Simplified)
    SimSun-ExtB Regular (Han Simplified)
    UD Digi Kyokashko N-B Bold (Japanese?)
    UD Digi Kyokashko NK-B Bold (Japanese?)
    UD Digi Kyokashko NK-R Regular (Japanese?)
    UD Digi Kyokashko NP-B Bold (Japanese?)
    UD Digi Kyokashko NP-R Regular (Japanese?)
    UD Digi Kyokashko N-R Regular (Japanese?)
    Yu Gothic (Japanese)
    Yu Gothic UI (Japanese)
    Yu Mincho (Japanese)
  • http://blog.adrianroselli.com/2015/01/on-use-of-lang-attribute.html On Use of the Lang Attribute
  • http://www.w3.org/International/questions/qa-lang-why Why use the lang attribute?
    en, ko, zh-Hant, zh-Hans, jp variations of a character (Unicode not known)
    use radical dictionary at https://www.mdbg.net/chinese/dictionary?page=radicals / top radical is rain / bottom makes snow.
    雪 (conventional font)
    Test added to examples below.
  • https://meiert.com/en/blog/20140825/html-and-language/ HTML and Specifying Language
  • https://medium.com/behancetech/localization-gotchas-for-asian-languages-cjk-e52a57c0fde1 Localization Gotchas for Asian Languages (CJK)
    Character variations 9aa8 66dc (Only getting 1 alternative font. How can I determine which font family is used?)
    Chinese simplified and traditional are messed up. cannot find appropriate radical simplifications/decorations.
    骨 曜 雪 雨 关 關 (conventional font)
    骨 曜 雪 雨 关 關 Simplified Chinese (Microsoft YaHei* need to verify font)
    骨 曜 雪 雨 关 關 Simplified Chinese (NSimSun Regular*; need to verify font)
    骨 曜 雪 雨 关 關 Simplified Chinese (NSimSun)
    骨 曜 雪 雨 关 關 Simplified Chinese (MS Hei)
    骨 曜 雪 雨 关 關 Simplified Chinese (MS Song)
    骨 曜 雪 雨 关 關 Simplified Chinese (SimHei)
    骨 曜 雪 雨 关 關 Traditional Chinese (Microsoft MHei* (guess); need to verify font)
    骨 曜 雪 雨 关 關 Traditional Chinese (MingLiU; need to verify font)
    骨 曜 雪 雨 关 關 Traditional Chinese (MingLiU-ExtB Regular*; need to verify font)
    骨 曜 雪 雨 关 關 Traditional Chinese (MingLiU-ExtB; need to verify font)
    骨 曜 雪 雨 关 關 Traditional Chinese (MingLiU_HKSCS; need to verify font)
    骨 曜 雪 雨 关 關 Traditional Chinese (Microsoft JhengHei; need to verify font)
    骨 曜 雪 雨 关 關 Japanese (MS Gothic Regular; need to verify font)
    骨 曜 雪 雨 关 關 Japanese (MS Gothic; need to verify font)
    骨 曜 雪 雨 关 關 Japanese (MS PMincho; need to verify font)
    骨 曜 雪 雨 关 關 Japanese (Meiryou; need to verify font)
    骨 曜 雪 雨 关 關 Korean (Microsoft NeoGothic (guess))
    骨 曜 雪 雨 关 關 Korean (Malgun Gothic (guess))
    骨 曜 雪 雨 关 關 Korean (Batang)
  • https://www.smashingmagazine.com/2012/07/12-commandments-software-localization/ 12 Commandments of Software Localization, Zack Grossbart

XLIFF[edit]

XLIFF localization.

Illustrations can avoid the segmentation step because SVG strings will usually be short, but title and desc elements may be exceptions.

There's a master file (say in .en). Then it is processed via .en.de XLIFF. Also a .en.es XLIFF.

There can be IDs, but they should be unique to the file. The lookup by ID. I want IDs that identify the content rather than the instance.

XLIFF filter: Take multilanguage SVG and convert to XLIFF. The trivial version is just an XSLT that grabs the text nodes. Slightly better version would pay attention to xml:lang, its:translate, and ITS rules. ITS rules could have "switch/text" and "@systemLanguage" awareness. The "@systemLanguage" attribute is further complicated because it is a list rather than just one IETF language code; see if XSLT can match substring patterns.

Need a second filter that produces the skeleton, and a third filter to merge XLIFF to language version.

Internationalization Tag Set[edit]

Interaction of XLIFF and other schemes with Internationalization Tag Set.

  • xml:lang attribute. HTML has compatible lang attribute that has consistency rules (if both exist, they must be the same).
  • its:translate attribute. HTML 5 has compatible global attribute translate.
  • its:locNote, its:locNotePointer, its:locNoteRef, its:locNoteRefPointer
  • its:term attribute. HTML 5 does not have this attribute. (http://www.w3schools.com/tags/ref_attributes.asp)
    global its:termRule requires one of its:termInfoRef (URI), its:termInfoPointer (XPath), or its:termInfoRefPointer (XPath)
    local its:term may have optional its:termInfoRef but not others
    its:termConfidence 2.0

Question: does SVG 2.0 have translate attribute? Apparently not.

There are data-* attributes in HTML 5 and SVG 2.

There are its-... attributes in HTML 5 (or at least the Nu validator). It makes local notes trivial. HTML with those attributes (camel-case conversion hack) will validate in Nu. 

<!DOCTYPE html>
<html lang=en>
  <head>
    <meta charset=utf-8>
    <title>Terminology test: default</title>
  </head>
  <body>
    <p>We need a new <span its-term=yes>motherboard</span>
    </p>
  </body>
</html>

I don't get the same luck when the its namespace is introduced. The Nu validator complains about introducing the new namespace and the new attribute. Of course, the validator does not have a specification for ITS. 

<!DOCTYPE html>
<html lang="en" xmlns:its="http://www.w3.org/2005/11/its" its:version="2.0" >
  <head>
    <meta charset=utf-8>
    <title>Terminology test: default</title>
  </head>
  <body>
    <p>We need a new <span its-term=yes>motherboard</span>
    </p>
  </body>
</html>

Now try with SVG. Augment the SVG 1.1 DTD to include some ITS definition. Rules are stuffed in metadata element. Local markup stuffed everywhere, but it could be tighter. 

<?xml version="1.0" encoding="utf-8" standalone="no" ?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[

<!-- Miniature DTD for ITS -->
<!ELEMENT its:rules         (its:termRule
                           | its:translateRule
                           | its:withinTextRule
                           | its:locNoteRule
                           | its:domainRule
                           )* >
<!ATTLIST its:rules         version       CDATA #REQUIRED
                            queryLanguage CDATA #FIXED "xpath"
                            xlink:href    CDATA #IMPLIED
                            xlink:type    (simple) #IMPLIED >

<!ELEMENT its:termRule      EMPTY >
<!ATTLIST its:termRule      selector  CDATA #REQUIRED >

<!ELEMENT its:translateRule EMPTY >
<!ATTLIST its:translateRule selector  CDATA      #REQUIRED
                            translate (yes | no) #REQUIRED >

<!ELEMENT its:withinTextRule EMPTY >
<!ATTLIST its:withinTextRule selector   CDATA      #REQUIRED
                             withinText (yes | no) #REQUIRED >

<!ELEMENT its:locNoteRule    (its:locNote)? >
<!ATTLIST its:locNoteRule    selector    CDATA #REQUIRED
                             locNoteType (description | alert) #REQUIRED 
                             locNotePointer    CDATA #IMPLIED
                             locNoteRef        CDATA #IMPLIED
                             locNoteRefPointer CDATA #IMPLIED >

<!ELEMENT its:locNote        (#PCDATA) >

<!ELEMENT its:domainRule EMPTY >
<!ATTLIST its:domainRule     selector CDATA #REQUIRED
                             domainPointer CDATA #REQUIRED
                             domainMapping CDATA #IMPLIED >

<!-- Include those elements in the SVG external subset -->
<!ENTITY % SVG.metadata.extra.content "| its:rules" >
<!ENTITY % SVG.External.extra.attrib "
  its:version   CDATA      #IMPLIED
  its:term      (yes | no) #IMPLIED
  its:termInfoRef CDATA    #IMPLIED
  its:termConfidence CDATA #IMPLIED
  its:translate (yes | no) #IMPLIED

  its:locNote     CDATA                 #IMPLIED
  its:locNoteRef  CDATA                 #IMPLIED
  its:locNoteType (description | alert) 'description'

  its:withinText  (yes | no | nested)   'no'
  ">
]>
<svg version="1.1"
     viewBox="0 0 640 320"
     xml:lang="en"
     xmlns:xlink="http://www.w3.org/1999/xlink"
     xmlns:its="http://www.w3.org/2005/11/its"
     its:version="2.0" >
  <title>Terminology test: default</title>
  <metadata>
    <its:rules version="2.0">
      <its:translateRule translate="no" selector="//term" />
      <its:withinTextRule withinText="yes" selector="//tspan" />
    </its:rules>
  </metadata>
  <text x="10" y="10" its:translate="yes">We need a new <tspan its:term="yes">motherboard</tspan>.</text>
</svg>

For SVG 1.2, there may be some ITS accommodations.

ITS 1.0 local markup

  • att.local.with-ns.attributes ::= att.local.with-ns.attribute.translate, att.local.with-ns.attribute.locNote, att.local.with-ns.attribute.locNoteType, att.local.with-ns.attribute.locNoteRef, att.local.with-ns.attribute.termInfoRef, att.local.with-ns.attribute.term, att.local.with-ns.attribute.dir

Quotation from the ITS v2.0 specification:

ITS 2.0 provides the following data categories:
  • Translate: expresses information about whether a selected piece of content is intended for translation or not.
  • Localization Note: communicates notes to localizers about a particular item of content.
  • Terminology: marks terms and optionally associates them with information, such as definitions or references to a term data base.
  • Directionality: specifies the base writing direction of blocks, embeddings and overrides for the Unicode bidirectional algorithm.
  • Language Information: expresses the language of a given piece of content.
  • Elements Within Text: expresses how content of an element is related to the text flow (constitutes its own segment like paragraphs, is part of a segment like emphasis marker etc.).
  • Domain: identifies the topic or subject of the annotated content for translation-related applications.
  • Text Analysis: annotates content with lexical or conceptual information (e.g. for the purpose of contextual disambiguation).
  • Locale Filter: specifies that a piece of content is only applicable to certain locales.
  • Provenance: communicates the identity of agents that have been involved processing content.
  • External Resource: indicates reference points in a resource outside the document that need to be considered during localization or translation. Examples of such resources are external images and audio or video files.
  • Target Pointer: associates the markup node of a given source content (i.e. the content to be translated) and the markup node of its corresponding target content (i.e. the source content translated into a given target language). This is relevant for formats that hold the same content in different languages inside a single document.
  • Id Value: identifies a value that can be used as unique identifier for a given part of the content.
  • Preserve Space: indicates how whitespace is to be handled in content.
  • Localization Quality Issue: describes the nature and severity of an error detected during a language-oriented quality assurance (QA) process.
  • Localization Quality Rating: expresses an overall measurement of the localization quality of a document or an item in a document.
  • MT Confidence: indicates the confidence that MT systems provide about their translation.
  • Allowed Characters: specifies the characters that are permitted in a given piece of content.
  • Storage Size: specifies the maximum storage size of a given piece of content.

Local Data Categories. There are some distinctions in ITS v2.0 when used with HTML. HTML has local markup for the Translate, Directionality, and Language Information data categories.

Translation Memory eXchange[edit]

Translation Memory eXchange is another XML format for translated pairs. Google Translate apparently accepts it. GT was free, but apparently was abused and is now a paid service.

  1. http://www.ibm.com/developerworks/xml/library/x-localis/
    TBX (TermBase eXchange) should be here. Has categories.
    "Fortunately, the XLIFF standard includes attributes for specifying string position in a dialogue, font type and size used for the text, and many other details. A tool specifically designed for software localisation can be used to visually adjust the dialogue layout during translation."
    "Java technology is the most widely used language after C/C++, and it is designed to facilitate software internationalization. Java applications that are ready for localisation store GUI text in resource bundles, plain text files that, like their C/C++ counterparts, make text extraction a simple task. However, Java technology has an advantage over C/C++: GUI objects can be dynamically adjusted at execution time according to the length of the text that the object contains. This feature makes the Java language one of the better choices for developing multilingual applications."
  2. http://www.ibm.com/developerworks/xml/library/x-localis2/index.html
  3. http://www.ibm.com/developerworks/xml/library/x-localis3/
    TMX is inadequate for glossaries (the word "gloss"!); TBL is better; can transform to each other
  • http://mymemory.translated.net/doc/spec.php
    http://api.mymemory.translated.net/get?q=stripper&langpair=en%7Cde&subj=Mechanical (JSON format) 
    {"responseData":{"translatedText":"Blockstripper","match":1},"responseDetails":"","responseStatus":200,"responderId":"235",
"matches":[
{"id":"441919990","segment":"stripper","translation":"Blockstripper","quality":"55","reference":" |@| ","usage-count":3,"subject":"Mechanical","created-by":"IATE","last-updated-by":"IATE","create-date":"2014-11-13 21:25:33","last-update-date":"2014-11-13 21:25:33","tm_properties":"[]","match":1},
{"id":"449281594","segment":"Stripper.","translation":"Vorrichtung zum Entfernen des Stanzabfalls.","quality":"55","reference":"http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/|@|http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/","usage-count":1,"subject":"Legal_and_Notarial","created-by":"Uni-heidelberg.de","last-updated-by":"Uni-heidelberg.de","create-date":"2014-11-28 17:25:00","last-update-date":"2014-11-28 17:25:00","tm_properties":"[]","match":0.98},
{"id":"451341627","segment":"Stripper","translation":"Abstreifeinrichtung","quality":"55","reference":"http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/|@|http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/","usage-count":1,"subject":"Legal_and_Notarial","created-by":"Uni-heidelberg.de","last-updated-by":"Uni-heidelberg.de","create-date":"2014-11-28 17:59:35","last-update-date":"2014-11-28 17:59:35","tm_properties":"[]","match":0.97}]}
    http://api.mymemory.translated.net/get?q=stripper&langpair=en%7Cde&of=tmx&subj=Mechanical (TMX format) 
    <?xml version="1.0" ?>
<!DOCTYPE tmx SYSTEM "tmx11.dtd">
<tmx version="1.1">
<header
  creationtool="MyMemory - mymemory.translated.net"
  creationtoolversion="Beta"
  segtype="sentence"
  o-tmf="TW4Win 2.0 Format"
  adminlang="EN-US"
  srclang="EN-GB"
  datatype="rtf"
  creationdate="20160929T211941Z"
  creationid="Translated">
  </header>
<body>
<tu tuid="0" creationdate="20160929T211941Z" creationid="IATE" usagecount="3" lastusagedate="20160929T211941Z">
	<tuv lang="EN-GB"  creationdate="20160929T211941Z" creationid="IATE">
	  <seg>stripper</seg></tuv>
	<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="IATE" changedate="20141113T212533Z" changeid="IATE">
	  <seg>Blockstripper</seg></tuv>
	</tu>
<tu tuid="1" creationdate="20160929T211941Z" creationid="MyMemoryLoader" usagecount="2" lastusagedate="20160929T211941Z">
	<tuv lang="EN-GB"  creationdate="20160929T211941Z" creationid="MyMemoryLoader">
	  <seg>Stripper.</seg></tuv>
	<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="MyMemoryLoader" changedate="20141128T172500Z" changeid="MyMemoryLoader">
	  <seg>Vorrichtung zum Entfernen des Stanzabfalls.</seg></tuv>
	</tu>
<tu tuid="2" creationdate="20160929T211941Z" creationid="MyMemoryLoader" usagecount="2" lastusagedate="20160929T211941Z">
	<tuv lang="EN-GB"  creationdate="20160929T211941Z" creationid="MyMemoryLoader">
	  <seg>Stripper</seg></tuv>
	<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="MyMemoryLoader" changedate="20141128T175935Z" changeid="MyMemoryLoader">
	  <seg>Abstreifeinrichtung</seg></tuv>
	</tu>
</body>
</tmx>

Segmentation Rules eXchange[edit]

Also Segmentation Rules eXchange.

SVG diagram translation[edit]

Consider doing XML, SVG, and XLIFF stunts with multiple files and with Wiktionary.

Editor provides single language PNG file. Bitmaps are not scalable. They are difficult to manipulate. For other wikis, translation is needed. Translations require erasure and addition.

Editor provides single language SVG file. Bare minimum: quick and dirty word substitution. Don't have sentences.

Where is the link for the tool that edits SVG files to make new language versions?

Worked on files[edit]

Some to do

Example images and issues[edit]

Consider the Gibraltar SVG file (a featured picture). Large SVG file, now many versions, and little ability to maintain it.

Derived from

Another map file:

An odd map file:

Dual-language Montreal wants labels in French and English, but a translation would only need one language. The file also has embedded raster graphics.

Other maps

Compare to some internally translated SVG files. Sadly, these files are generated by hand. Pulling them into an SVG editor can strip the translations.

Here's a hand-edited file that has included translations; there are 49 different IETF language tags in the file. The file also uses some SVG context sensitive features that are lost in the PNG translation.

Mediawiki File inclusion mechanics[edit]

How does the wiki access different versions? It tacks on &lang=IETF to the link. For example, the line

  • [[file:First Ionization Energy.svg|lang=de|thumb|left|German version]]

Inserts this image:

German version

by inserting the following HTML: 

<div class="thumb tleft">
<div class="thumbinner" style="width:222px;"><a href="/w/index.php?title=File:First_Ionization_Energy.svg&amp;lang=de" class="image"><img alt="" src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-220px-First_Ionization_Energy.svg.png" width="220" height="92" class="thumbimage" srcset="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-330px-First_Ionization_Energy.svg.png 1.5x, //upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-440px-First_Ionization_Energy.svg.png 2x" data-file-width="1200" data-file-height="500" /></a>
<div class="thumbcaption">
<div class="magnify"><a href="/wiki/File:First_Ionization_Energy.svg" class="internal" title="Enlarge"></a></div>
German version</div>
</div>
</div>

The anchor element makes a click go to the file page with the appropriate language selected for the SVG. The img element displays the picture. Not sure what 1/1d means, but it may be a cache issue. Image directory has lang(IETF langtag)-(size)px-(name).png made by renderer. The srcset attribute provides magnify/"enlarge" options? Notice data-file-width and data-file-height attributes that may anticipate problems with SVG files containing hard widths and heights.

Check langtag issue at Phab:T154237. librsvg does not handle hyphenated langtags correctly, so zh-hans and zh-hant just match to zh and look the same:

Hans version
Hant version

Mediawiki hypenated langtag stopped working[edit]

In September 2017, I noticed the above two images (zh-hans an zh-hant) were not working at all (not just selecting the wrong version). Try just zh:

zh version

New development of Phab:T154132.

So first string works, but second does not.

  • src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langzh-220px-First_Ionization_Energy.svg.png"
  • src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langzh-hans-220px-First_Ionization_Energy.svg.png"

Guessing that parameter extraction is now confused by hyphenated langtags.

Lazy evaluation suggest the URL is grabbed at .../thumb/... and then parsed

Hypenated language tag problem fixe 26 October 2017 by Gilles. UI problems still present.

Automatic use of IETF langtag[edit]

Want each wiki to automatically use the language tag. Somewhere there was a comment that this will come.

The periodic table just needs translations for "group" and "period".

Here's a file with just 6 sentences to translate.

Interesting problem with language variation (accent variation between pt-BR and pt-PT: Hidrogênio and Hidrogénio):

Also look forward to direct insertion of SVG files rather than the current convert to PNG.

Here's a JPEG file that is used on many projects and that people want translated. An SVG version without English captions is also desired.

A simpler diagram

Needing conversion; spot labels

Several images use numbers, but have translations on their description pages.

Planetary[edit]

Chemistry see also[edit]

file not needing translation (but conversion to SVG)

2019 Wishlist survey to add Chemical File Format

Anatomical diagrams[edit]

Lots of diagrams with minimal text changes.

This file has some embedded fonts.... Complicated shading.

Ripe files[edit]

Good targets are simple SVG files that are used on several foreign WPs. If the file isn't used, translation is not a priority.

May want to use {{translation possible}} |switch=value for many other uses. ITS tags, for example.

Matching numbers and units (e.g., 30 volts).

Good candidates (see Commons:Category:Translation possible - SVG (switch))

Dubious SVG:

Switch-translated files that don't get much use. Note temperature scale/labeling issues.

Some files can be problematic even if converted to SVG (e.g., dense diagram, fr translations overlong)

SVG translation methods[edit]

No translation[edit]

Some files do not need translation because they do not use words. Many diagrams use numbers to identity components.

A benefit to this method is the diagram is universal. It also does not have problems with multi-line labels.

The disadvantage is the translation must be done each time the image is used: the caption must explain what the numbers mean. In other words, the image does not stand by itself but rather requires an external explanation.

Multi-file translation[edit]

In this method, an existing file is edited by replacing the existing text with translations and then saved under a new name. It is the default for Wikimedia Commons because it is easy to do. Pull the existing file into Inkscape, edit it, and save it.

The problem is it makes many copies of essentially the same file. If someone wants to change the diagram a little or change the color scheme, it must be done for each copy.

A better method uses one copy of the file. There are two methods: switch and XLIFF. Wikimedia Commons is not set up for XLIFF.

Switch translation[edit]

Better idea about how Commons approaches the translation issue.

The translation markup.

In an SVG file, find text blocks or spans that permit translation. 

<text x="100" y="200">House</text>

converts to 

<g transform="translate(100,200)">
  <switch>
    <text systemLanguage="de">Haus</text>
    <text>House</text>
    </switch>
  </g>

or even 

<switch transform="translate(100,200)">
  <text systemLanguage="de">Haus</text>
  <text>House</text>
  </switch>

The addition of switch has a small impact (roughly 16 bytes). Each clause costs at least 33 bytes plus the translation bytes. Estimate the translation is 17 bytes. Then each language adds 50 bytes. With fifty languages, a single term adds 2.5 kB. Forty terms would add 100 kB to an SVG file. That relies on the position information being hoisted so it is not repeated for each translation. File:Map of USA with state names.svg paid 400 kB even though it has hoisted location information; I don't know where the penalty is; 160 bytes per language term seems high.

For ordinary SVG files, the systemLanguage is compared to the user's preferences. SVG 1.1 has slightly odd semantics for choosing a preference, but that is being straightened out in SVG 2.0.

WP throws a further wrench in the preference calculation because it does not serve the SVG but rather a bitmap. That bitmap is built by rsvg which only takes a single language. The lang argument should be a preference list. If the illustration does not have a desired language, then "en" may not be the best fallback. An "sr-Cyrl" might better fallback to "sr-Latn" and after that "ru". It's a minor point, and fixing it might be too much trouble. Serving the switch SVG would fix it. A German reading the de.WP might want the fallback to be "en", but a Swiss might want the fallbacks to be "fr" and "it" before "en".

A problem with serving SVG is there are some very piggish SVG files. A 200 px PNG thumbnail has a bounded size.

Consider problems when the text has leaders that point to places in the image.

Problem with expansion. Some languages have succinct labels, but others require more space. Don't consider problem when line breaks are needed. That really is an XLIFF issue.

Flowed text not available/may not help
see also C:Help:SVG#Black rectangle (Flowed Text bug)

There are things that should not be translated. Some phrases are intended to stay in a specific language; a diagram that explains translation, for example, would keep the native language. Names are often not translated (Casablanca does not become White House), but they may be Anglicized (e.g., Чебышёв becomes Chebyshev). Chemical symbols are not translated. Symbols may change (input voltage V becomes input U). What about the subscripts?

Conjugation and declension. Would not expect a dictionary to have "red house" or "three houses". Google Translate and some matching?

Display multiple outlines -- enough room so everything fits. This could make diagrams in some languages look poor.

There is an issue with mixins and validation.

There is an issue with <title>, <desc>, and <hint> elements because they are not switched. Conditional processing applies to display elements, but not others such as titles and descriptions. Consequently, we cannot do this: 

<svg ...>
  <switch>
    <title systemLanguage="de">Wien</title>
    <title systemLanguage="en">Vienna</title>
    <title>Vienna</title>
    </switch>
  </svg>

That means it will be difficult or impossible to translate titles/tooltips. On the other hand, XLIFF would have no trouble with those elements.

SVG 2.0 working draft recognizes this problem,[1] but does not use systemLanguage. Instead, title and desc may appear multiple times with different lang attributes (which are single valued). The construct (which is not yet supported in browsers: Issue 67) is 

<svg lang=en...>
  <title>Vienna</title>
  <title lang="de">Wien</title>
  <title lang="ru">Вена</title>
  ...
  </svg>

More exotic issues: mouseover. I believe this fails and is better handled with XLIFF or separate rectangles. 

<rect x="100" y="200" w="20" h="40">
  <title>House</title>
  <title lang="de">Haus</title>
  <desc>Blue</desc>
  <desc lang="de">Blau</desc>
  </g>

Planar translation[edit]

Several files do not keep the translations together but rather separate languages into a planar configuration. Consequently, there is only one switch element and only one clause per language. Not repeating the language selection clausest suggests a byte savings, but each text element must repeat the position information — something whose cost would match the cost of repeating systemLanguage="xx". In addition, the planar configuration does not keep the translated terms together. Any equivalence must be inferred from the similarity in position in the diagram or order within the clauses. 

<switch>
  <g systemLanguage="de">
    <text x="100" y="100">Strasse</text>
    <text x="100" y="200">Haus</text>
    </g>
  <g>
    <text x="100" y="100">Street</text>
    <text x="100" y="200">House</text>
  </switch>

In addition, there are different methods of planar association. A plane might be a g element with text children, or it might be a text element with tspan children. The latter method complicates multiline phrases. If "Atlantic Ocean" needs to be broken into two lines, it is better if the lines share a unique parent so they hang together. A simple view is each phrase should exist in its own text element; if the phrase is broken over lines, then those lines are tspan elements within the text element. Otherwise, all the words are just shuffled together.

I would deprecate the planar method, but Graphics Lab editors seem to favor it.

  • File:Epicenter Diagram.svg (switch-text-tspan)
  • File:Galaxies-AGN-Inner-Structure.svg (switch-g-text-(tspan for multiline))
    This diagram gets even screwier because simple translations are called out in a full matrix
    Non-Unicode fonts are font-family="Helvetica" (but it is not a style!)
    Chinese fonts are font-family="Noto Sans CJK TC" (but it is not a style!)
    font-size=10 most of the time, but font-size=12 for title
    fill=white because on black background
    one line is split, so it uses tspan; otherwise tspan is not used.
    x,y are usually called out as 0 (because transform is there); tspan uses nonzero y
    that means simple x,y cannot be used to identify translation groups; must invert the transform or use the transform!
    switch languages improperly ordered as zh, zh-CN, zh-TW; therefore only first zh controls; zh-Hans and zh-Hant would be better
    apparently sequence is zh-Hans, zh-Hans, zh-Hant...
    see also Chinese Wikipedia for discussion of zh-Hans, zh-Hant, zh-CN, zh-TW, zh-SG, zh-HK, zh-MO
    see also meta:Automatic conversion between simplified and traditional Chinese
  • File:Multilingual SVG example.svg (switch-text-tspan) (but only 1 item to translate)
    has screwy dy="-5" for first line.

Here's a switch file that has tspan elements to do multiline translations.

Other switch problems[edit]

If userLang=ec, we take the first. If userLang=ec-EL, we bypass the first.

Many SVG files have language order problems. (Is there a allowReorder section?) For example, this ordering is wrong because SVG 1.1 rules hide the following sr-EC and sr-EL langtags: 

<text systemLanguage="sr">Актиноид</text>
<text systemLanguage="sr-EC">Актиноид</text>
<text systemLanguage="sr-EL">Aktinoid</text>

A better order is 

<text systemLanguage="sr-EC">Актиноид</text>
<text systemLanguage="sr-EL">Aktinoid</text>

The reorderings have slight problems because "sr-EC;q=1.0,sr-EL;q=0.8,sr;q=0.5" and "sr-EL;q=1.0,sr-EC;q=0.8,sr;q=0.5" will always choose the first option under SVG 1.1 rules.... Maybe the proper viewpoint is the undesired script must be set below the "sr" q value.

Reasonable sorting algorithm? Since systemLanguage can take many langtags, a reordering cannot solve the problem in general.

Here's a single-switch/multi-switch file that put Ukrainian translation after the default translation (uk translation now reordered). Fault line was incorrect (should be fault plane). Only a few terms to translate: epicenter, fault plane, hypocenter, and focus. Wiktionary gives good coverage for British epicenter, modest for focus, and poor for the rest. File has funky rotation using a transform matrix despite line drawn elsewhere.

A file with no default language:

  • File:Psychrometer2.svg no default language!
    also horrible typesetting for non-switch formula ΔT = Ttr - Tf with colors

Text on a path[edit]

SVG renderers may not support all SVG features. For example, rsvg does not support textpath; consequently, many users take the text and convert it to curves.

Switch features (requiredFeatures attribute and hasFeature DOM)

The feature#Text should signal that textpath is not available. rsvg does not claim feature#Text:

However, some browsers may claim the feature but not do all the alignments; that may be inappropriate.

  • Edge claims, does anchor but not baseline alignment
  • Firefox does not claim, does anchor but not baseline alignment
  • rsvg does not claim, does anchor but not baseline alignment
  • Chrome ...

Does rsvg have partial support? 

<switch>
  <textpath requiredFeatures="http://www.w3.org/TR/SVG11/feature#Text" ...> ... </textpath>
  <path ...> ... text as curves ... </path>
  </switch>

There's also the text baseline alignment mess to figure out. dy can be used, but there are also some CSS modifications that can be done. It may be that CSS baseline alignment works when baseline not available, but it seems misused.

  • vertical-align (applies to element):
    top
    bottom
    text-top
    text-bottom
    baseline
    middle
    sub
    super
  • Trial linesublinesuperline.

There's also SVG 1.2 textArea with its feature:

SVG Optimizer[edit]

SVG Optimizer additions?

SVG direct (HTML 5)[edit]

Mediawiki blocks some elements....

Trying direct SVG inclusion (blocked): 

 <svg xmlns="http://www.w3.org/2000/svg" width="320" height="160">
    <rect x="0" y="0" width="320" height="160" fill="cyan" />
    </svg>

Using tag magick worter...

<svg xmlns="http://www.w3.org/2000/svg" width="320" height="160"><rect x="0" y="0" width="320" height="160"></rect></svg>

Try object: 

 <object width="320" height="160"
         data="https://upload.wikimedia.org/wikipedia/commons/1/1d/First_Ionization_Energy.svg"
         type="image/svg+xml" >
   </object>

Trying bdi: bdi and bdo. These elements work.

Translation sources[edit]

User translation[edit]

User could supply translations as in the current SVG Translate tool.

We don't get a lot of languages at once. Want the image to explode. User draws and SVG diagram and labels it. Diagram then becomes available with translations in many languages.

Google Translate[edit]

Good coverage.

Get a lot of languages, but theres a problem with getting the wrong sense of a word.

I don't know how to qualify the field to Google Translate, and the translations are probably better for long sentences rather than isolated words.

Just make it easy to correct the translations.

Wikidata[edit]

Conventional approach is a database of translations.

This could have field and tracked corrections. Require some subject indicator or word sense.

There's a difficult issue here. Image must be accurately labeled, but then the labels may be too fine (no translations) or too broad (word confusion still exists).

The field could be gleaned from some categories for the image.

That brings up a separate issue. Is the Gibraltar map present many times in a category? Better to have one image and select the right language depending on the user's preference.

Articles have interlanguage links, and they can be used to extract translations.

Wikidata access[edit]
  • WP:Wikidata#Inserting Wikidata values into Wikipedia articles
  • Category:Wikidata templates
  • {{wikidata}} exotic
  • {{wikidata list}} takes SPARQL...; do not use in mainspace
  • {{Q}}: Douglas Adams (Q42), Douglas Adams (Q42), ionization energy (P2260)
  • {{Label}}: Douglas Adams, has part(s)
  • {{wikidata entity}}: Douglas Adams (Q42), ionization energy (P2260)
  • {{wikidata entity link}}: Douglas Adams (Q42), ionization energy (P2260)
  • {{wikidata property link}}: ionization energy (P2260), ionization energy (P2260)
  • {{wikidata}}
    • Dianne Feinstein (Q230733) position held (P39)
      strange: California (Q99) has a property head of government (P6) to get the governor (Gavin Newsom Edit this on Wikidata),
      but its senators must be found indirectly with a query.
      properties: Mayor of San Francisco, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator
      properties (current):
      properties (qualifier electoral district (P768)): Mayor of San Francisco, United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat), United States senator (California Class 1 senate seat)
      properties (qualifier start time (P580) end time (P582)): Mayor of San Francisco (4 December 1978; 8 January 1988), United States senator (4 November 1992; 3 January 1993), United States senator (3 January 1993; 3 January 1995), United States senator (3 January 1995; 3 January 1997), United States senator (3 January 1997; 3 January 1999), United States senator (3 January 1999; 3 January 2001), United States senator (3 January 2001; 3 January 2003), United States senator (3 January 2003; 3 January 2005), United States senator (3 January 2005; 3 January 2007), United States senator (3 January 2007; 3 January 2009), United States senator (3 January 2009; 3 January 2011), United States senator (3 January 2011; 3 January 2013), United States senator (3 January 2013; 3 January 2015), United States senator (3 January 2015; 3 January 2017), United States senator (3 January 2017; 3 January 2019), United States senator (3 January 2019; 3 January 2021), United States senator (3 January 2021; 3 January 2023), United States senator (3 January 2023; 29 September 2023)
      properties (qualifier replaces (P1365)): Mayor of San Francisco (George Moscone), United States senator (John Seymour), United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator
      properties (qualifier series ordinal (P1545)): Mayor of San Francisco (38), United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator, United States senator
    • helium (Q560) mass (P2067)
      properties: 4.002602 dalton
      properties (raw): 4.002602
    • Utrecht (province)

The image File:BirdBeaksA.svg has some interesting issues. It is not about the actual birds but rather their beaks. The Commons page also has interesting links.

  • Take lesser yellow-headed vulture (Q852751).
    The species is taxon name (P225): Cathartes burrovianus
    What are other ways to do select German name? lesser yellow-headed vulture (Q852751) (not working)
    Aliases: Cathartes burrovianus, savannah vulture
    There are aliases, but there is also taxon common name (P1843): Lesser Yellow-headed Vulture
    Languages
    Multilanguage: fwltur penfelyn bach, Lesser Yellow-headed Vulture, Kleiner Gelbkopfgeier, Aura sabanera, zopilot de sabana, kondor menší, rohtlakondor, Myrkondor, Малая желтоголовая катарта, キガシラコンドル, Gulhovedet Grib, pikkukondori, kis sárgafejű keselyű, sępnik pstrogłowy, kondor modrohlavý, savanngam, Kleine Geelkopgier, Urubu à tête jaune, Avvoltoio testagialla minore, urubu-de-cabeça-amarela, mažasis kalakutinis grifas, Jote cabeza amarilla, النَّسر الأصغر أصفر الرأس, катарта саванова, Savannen-Gelbkopfgeier, mali lešinar
    How does one select the German common name?
    Lesser Yellow-headed Vulture ("language" qualifier does not work?)
    Lesser Yellow-headed Vulture (list qualifiers fails)
    Link to commons category with Commons category (P373): Commons:Category:Cathartes burrovianus
Wikidata map issues[edit]

Maps often represent a point in time.

So Q item current versus another time.

Chennai: Chennai (Q1352), Madras in aliases: Madras, official name (P1448) : Madras

Burma, Myanmar Myanmar (Q836), aliases: Burma, Republic of the Union of Myanmar, 🇲🇲, Union of Burma, MM, BUR, MYA, official name (P1448): Republic of the Union of Myanmar

St Petersberg, Petrograd, Leningrad. country (P17) changes a lot.

Wikidata orbits[edit]

Planet Nine search has many orbital maps in PNG. It's easy to draw an ellipse in SVG, but the coordinates need to be mapped. The Geo mapping probably does not cover that issue well. Project into the ecliptic.

Wikidata should provide the orbital elements, but that has not made it to Wikidata yet. See (496315) 2013 GP136 (Q22231120), which has no orbital elements. Compare to (496315) 2013 GP136 which has the elements.

Wikidata could be a star catalog. Cygnus (constellation).

Wikidata API[edit]

If I know the Q number for a list (eq Douglas Adams (Q42)), then (format= json, jsonfm, xml, xmlfm)

Look at some items.

Problems...

This gives several Qs

API definition

Component[edit]

An object may have a WD item but its components may not.

Consider an illustration of a MOSFET. It may have drain, gate, source, channel, substrate, insulator, depletion region.

See illustrations at

Another series (with damaged illustrations)

Wikidata hierarchy does not have components.

Do have

Consider Wictionary....

SPARQL queries[edit]

Had trouble with earlier, but starting to get results: 

  • SELECT ?item { ?item p:P246 [] . }
    P246 is element symbol
    try it
  • SELECT ?item { ?item wdt:P31 wd:Q11344 . }
    P31 instance of Q11344 chemical element
    try it
  • Predecessor/Successor sometimes hang off of Instance of Chemical element and sometimes main item
    This problem has been fixed for most low elements but still exists in very high atomic numbers.
    P155 follows
    P156 followed by
  • works in low elements, but I think in wrong place for higher elements
    See rubidium (Q895) which has follows/followed by hanging off instance of chemical element (Q11344)
    See neptunium (Q1105) last hanging off instance of
    See lawrencium (Q1905) which has separate follows/followed by properties
  • SELECT ?item { ?item wdt:P31 wd:Q11344 . filter not exists { ?item p:P155 [] } }
    try it
    also do with "... filter exists ..." to get complement
  • try syntaxhighlight 
    SELECT ?item { ?item wdt:P31 wd:Q11344 .
               filter not exists { ?item p:P155 [] } }
  • That leaves open a CSS stunt where contents builds an href.
  • but there is {{SPARQL}}:
    {{wikidata:Template:SPARQL|query=SELECT ?item { ?item wdt:P31 wd:Q11344 . filter not exists { ?item p:P155 [] } } }}
  • Lots of elements (47 on 4 March 2017; 72 if drop the ?pred constraint) do not have mass information 
    SELECT ?item ?itemLabel ?mass ?atomnumber ?pred ?predLabel
{ 
  ?item wdt:P31 wd:Q11344 ;
        p:P31 ?stmt ;
        wdt:P1086 ?atomnumber . # data type is a quantity
  ?stmt pq:P155 ?pred .
  
  # not all elements have mass!
  filter not exists { ?item wdt:P2067 ?mass . } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .}
} ORDER BY ?atomnumber
    try it
  • filtered atomic number .le. 118 gives 35 elements with no mass entry 
    SELECT ?item ?itemLabel ?mass ?atomnumber
{ 
  ?item wdt:P31 wd:Q11344 ;
        wdt:P1086 ?atomnumber . # data type is a quantity
  
  # not all elements have mass!
  filter not exists { ?item wdt:P2067 ?mass . } .
  
  filter (?atomnumber <= 118) .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .}
} ORDER BY ?atomnumber
    try it
  • Look for elements that have an ionization energy. Desired units are eV, but sodium also has kcal/mol. 
    SELECT ?item ?itemLabel ?ie ?unitLabel ?ord ?atomnumber
{ 
  ?item wdt:P31 wd:Q11344 ;
        wdt:P1086 ?atomnumber .
  
  # not all elements have ionization energy
  { ?item p:P2260 ?stmt . 
    ?stmt pq:P1545 ?ord .
    ?stmt psv:P2260/wikibase:quantityAmount ?ie .
    ?stmt psv:P2260/wikibase:quantityUnit ?unit .
  } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .}
} ORDER BY ?atomnumber
    try it

I added mass for hydrogen, lithium, and nitrogen.

Push endpoint:

The wikidata entry for oxygen oxygen (Q629) gave O2 as an alias; I removed it because the article is about the element. There is a dioxygen (Q5203615) which is a simple substance but is not an instance of a gas. fluorine (Q650) is an instance of a gas. nitrogen (Q627) is not an instance of a gas. carbon (Q623) is not an instance of a solid. There's a logical problem here. An element is an element; it does not have to be in any phase. I need to distinguish state of matter (Q11430) and phase (Q104837), an aspect of state. Some elements form diatomic molecules, and those molecules can be solid, liquid, gas, plasma under certain conditions. Oxygen can be, for example, a liquid. Any statement about state/phase should contain conditions. But can just an element be something if it will form molecules? Copper copper (Q753)???

Political[edit]

On 21 August 2017, Wikidata sez there are 1,508 United States Senators who are still in office (only 1,504 if we insist the Senators be human; Wikidata knows 1,319 are dead): 

SELECT ?person ?personLabel ?stateLabel ?bdate ?ddate ?endtime
{ 
  ?person p:P39 ?stmt .
  ?person wdt:P569 ?bdate .
  ?stmt ps:P39 wd:Q13217683 .
  optional { ?person wdt:P570 ?ddate . } .
  optional { ?stmt pq:P768 ?state . } .
  filter not exists { ?stmt pq:P582 ?endtime } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .}
  
} ORDER BY ?stateLabel
try it

I did this query and spent a day pushing down to 100 entries: 

SELECT ?person ?personLabel ?stateLabel ?bdate ?sdate ?ddate
{ 
  ?person p:P39 ?stmt .
  ?person wdt:P569 ?bdate .
  ?person wdt:P31 wd:Q5 .
  ?stmt ps:P39 wd:Q13217683 .
  optional { ?stmt pq:P768 ?state . } .
  optional { ?stmt pq:P580 ?sdate . } .
  filter not exists { ?person wdt:P570 ?ddate . } .
  filter not exists { ?stmt pq:P582 ?endtime . } .
  filter not exists { ?stmt pq:P1366 ?succby . } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .} .
  
} ORDER BY ?stateLabel
try it
Political 2[edit]

I ran this query with several governor-of-state. VP Pence was still listed as governor of Indiana. 

SELECT ?item ?itemLabel ?ord ?startdate ?enddate ?successorLabel WHERE {
  ?item p:P39 ?stmt.
  ?stmt ps:P39 wd:Q16147601.
  OPTIONAL { ?stmt pq:P580 ?startdate. }
  OPTIONAL { ?stmt pq:P582 ?enddate. }
  OPTIONAL { ?stmt pq:P1366 ?successor. }
  OPTIONAL { ?stmt pq:P1545 ?ord. }
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
ORDER BY ?startdate

SELECT ?item ?itemLabel ?headLabel ?govLabel WHERE {
  ?item wdt:P31 wd:Q35657.
  optional { ?item wdt:P1313 ?head } .
  optional { ?item wdt:P6 ?gov } .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
ORDER BY ?itemLabel
Political 3[edit]

Fixed list of California Senators. 

SELECT ?person ?personLabel ?stateLabel ?bdate ?sdate ?endtime ?ddate
{ 
  ?person p:P39 ?stmt .
  ?person wdt:P569 ?bdate .
  ?person wdt:P31 wd:Q5 .
  ?stmt ps:P39 wd:Q13217683 .
  { ?stmt pq:P768 wd:Q99 . } . # ?state . } .
  { ?stmt pq:P580 ?sdate . } .
  OPTIONAL { ?person wdt:P570 ?ddate . } .
  OPTIONAL { ?stmt pq:P582 ?endtime . } .
  # filter not exists { ?stmt pq:P1366 ?succby . } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .} .
  
} ORDER BY ?sdate
Units[edit]

Different languages use different scripts, so the symbols used for those units may vary. 

SELECT ?item ?itemLabel ?unitLabel ?scriptLabel ?langLabel
{ 
  # BIND (wd:Q25406 AS ?item)
  # BIND (wd:Q83327 AS ?item)
  BIND (wd:Q11574 AS ?item)
  
  #find an item with a unit symbol
  ?item p:P558 ?stmt .
  
  #get the unit
  ?stmt ps:P558 ?unit.
  
  #get the script qualifier for the statement
  optional { ?stmt pq:P282 ?script . } .
  
  # get the language qualifier for the statement
  optional { ?stmt pq:P407 ?lang . } .
  
  SERVICE wikibase:label {
     bd:serviceParam wikibase:language "en" .}
}
try it
Songs[edit]

Songs should have a composer and a lyricist. (Compare instrumental Classical Gas (Q489275).)

John Lennon (Q1203) and Paul McCartney (Q2599) had a contract to be joint. Look at all songs performer (P175) The Beatles (Q1299). That should give an interesting result with respect to RDF bag or individual composer (P86) and lyrics by (P676). Some songs get hits for Lennon–McCartney (Q239177). 

SELECT ?song ?songLabel ?composer ?composerLabel ?lyricistLabel
WHERE
{
  ?song wdt:P31/wdt:P279* wd:Q7366 .
  ?song wdt:P175 wd:Q1299 .
  
  OPTIONAL { ?song wdt:P86 ?composer . }
  OPTIONAL { ?song wdt:P676 ?lyricist . }
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
ORDER BY ?song
try it
Typeface[edit]
SELECT ?typeface ?typefaceLabel
WHERE
{
  ?typeface wdt:P31/wdt:P279* wd:Q17451 .
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
ORDER BY ?typefaceLabel
try it

Wiktionary[edit]

.see d:Wikidata:Wiktionary

Diagrams are often associated with particular fields of study such as chemistry, physics, or mechanical engineering.

The Wiktionary translations are spotty (especially with technical terms). Gloss/distinctions are troublesome or even missing; compare wikt:stripper. Some common terms have good coverage. See, for example, wikt:radon, but notice Uzbek Latin and Cyrillic entries. Each entry is a span with class attribute of "Latn" or "Cyrl" and lang attribute of "uz".

In 2019 Wishlist Survey, there's a proposal to link Wikidata items to Wiktionary defs

In 2019 Wishlist Survey, there's a proposal to rationalize Wiktionary translations.

For reference:

Mechanical engineering

Diagram (which has been removed from Commons) has only one broken word....

English German
deep drawing Tiefziehen (deep pull)
workpiece Werkstück
blank Ronde
(not a place on a form)
blank holder Niederhalter (hold down)
die (piercing, blanking, drawing) Ziehmatrize (pull die)
drawing die Ziehmatrize
punch Ziehstempel (pull stamp)
(not what a boxer does)
stop
automatic stop
clearance (not the sale)
distance between two moving objects [sic]
die holder
stripper Abstreifer (=wiper/scraper)
(not the entertainer)
one who removes their clothes in a sexually provocative manner
chemical or tool to remove paint, sheathing
tool used to strip tubing
guide pins (plural!)


Physics

English terms
French terms
German terms
numbered

Diagram has lots of multiline statements. A better layout would use single-line labels, but it would look different. There is a mostly single-line version.

Mostly single-line version
English German
atomic number Ordnungszahl
Kernladungszahl
Atomnummer
glow discharge Glimmentladung
cathode fall (missing)
Cathode Kathode
https://de.wiktionary.org/wiki/Kathode
Aston dark space astonscher Dunkelraum (m)
cathode glow Kathodenglimmhaut
cathode dark space
Crookes dark space
Hittorf dark space (sp?)		 hittorfscher Dunkelraum (m)
Crookes
Cathode
negative glow negatives Glimmlicht (n)
Faraday dark space faradayscher Dunkelraum (m)
positive column positive Säule (f)
anode glow Anoden-Glimmlicht
anodisches Glimmlicht (n)
anode dark space anoden-Dunkelraum
anodischer Dunkelraum (m)
anode Anode

Unfortunately, Wiktionary does not have good coverage of the terms. I had put many of the physics terms in some time ago.

Would want a program that updates a diagram. If new translations are added, then they should be incorporated.

Pottery[edit]

DOM processing[edit]

For UI, what hacks can I do with DOM?

SVG authoring[edit]

Watt makes many points.

  • Don't include a DOCTYPE declaration in the SVG. A DTD was not written for SVG 1.2.
  • Avoid using style= because that confuses content and presentation:
    Usually this is simply gratuitous use of the style attribute, and something that should be avoided. Contrary to popular belief the style attribute (note, that's attribute, not element) doesn't separate content from presentation, and unless you need to override properties set by a CSS selector, it is best to use the SVG formatting attributes instead.
    Inkscape uses style all over the place.
    I would use style for capitalization hacks in maps: let the content be "Oregon" but the presentation be "OREGON".

Wiki's translation memory[edit]

Hey, there is a translation in the file namespace:

Which should relate to

Where srt is SubRip format. HTML 5 track element wants WebVTT (vtt). The translation might be faster to wait for .srt to change and then convert it to .vtt.

Look at how Commons serves the track information:
<track src="//commons.wikimedia.org/w/index.php?title=TimedText:The_GLAM-Wiki_Revolution.webm.de.srt&amp;action=raw&amp;ctype=text%2Fx-srt" kind="subtitles" type="text/x-srt" srclang="de" label="Deutsch (de) subtitles" data-mwtitle="TimedText:The_GLAM-Wiki_Revolution.webm.de.srt" data-mwprovider="local" data-dir="ltr"/>
The game is involved. Get the raw text but make sure it is served as text/x-srt. Somewhere there was a note that track required Content-Type.

How are these set up?

Issues...

Figure out MyLanguage hack. Is resolved during translation?

SVG Translate tool[edit]

There is an existing tool to aid SVG translations. See

The pages states, "This tool has been used to translate approximately 865 files since March 2011."

Just realized a confusing issue with this tool. X creates the original en version of an image. Y translates the image into fr; the Commons page suggests translating the file. Z comes along and translates fr version into de. We've translated a copy. Want that functionality (some translators won't know en), but the non-translated source should come from the original rather than the copy.

Apparently some of the translated files have been translated by pulling them into Inkscape and just editing the labels.

Architecture[edit]

SVG Translate is a PHP server-side tool.

It offers no translation help.

It throws up the text/tspan contents and asks the user to translate them.

Instead of using DOM processing, it uses pattern matching. The algorithm acknowledges that there will be problems for unusual element names. Stuffing a switch-translated SVG should have interesting results.

There is no attention to its: rules and attributes.

It copies the file and edits it. Consequently, there are now multiple copies (one per language) to maintain of the essentially the same information.

The industry route would be to have a skeleton file with XLIFF files to fill in the translations. Wiki are not set up to do that. There is an i18n file hierarchy for templates.

  • file
  • file/i18n/de
  • file/i18n/en
  • file/i18n/it
Throws an exception[edit]

When I tried to use it, it threw an exception.

Sometime later (ca. September 2016) it would throw an exception about Intuition::getLangNames()

OK, assuming svgtranslate from

is the beast, then 

	$html .= '<tr><th align="right">' . _html( 'th-language' ) . _g( 'colon-separator' ) . '</th><td>';
	$html .= "<select name=\"targetlanguage\" style=\"width: 40em\">\n";
	$langnames = $I18N->getLangNames();
	$default = $I18N->getLang();
	foreach( $langnames as $code => $name ){
		$html .= "\t<option value=\"$code\"";
		if( $code == $default ){
			$html .= " selected='selected'";
		}
		$html .= ">$name</option>\n";
	}

Intuition has published interfaces getLangName and getLangNames.

An edit in March changed public getLangNames to protected getLangNames on line 659 (now 661).

Code located in

Got phabricator wiki account and left comment about origin of SVG Translate bug.

Need GitHub account to complain about Intuition (but there may be a fork?).

I left some mail, and Krinkle fixed Intuition to make interface public again.

Another error - https
SVG Translate now working[edit]

Well, at least more parts are working. I never tried to save anything, but there is a report that saving the file does not work.

Sadly, the source JPEG gets more use on xx.WPs:

It is a standard SVG map that has many translations, but it has some translation issues.

It does not specify a lang in its svg element.

SVG does not validate. Issues Inkscape and RDF warnings. Then 4 errors for illegal characters in id string such as: 

id="texte &quot;Pays&quot;"

The problem is the normalized value of the xml:id should be an NCName, and NCNames should not have spaces or quotation marks. See http://www.w3.org/TR/xml-names11/#NT-NCName

Much of the file is graphics (extensive paths for borders) and internal bookkeeping (including worthless perspective defs); little is the actual text. It might be interesting to run svgo on it. Does svgo dissect the style attribute? Yes, see convertStyleToAttrs; moveElemsAttrsToGroup.

All capital names such as "SPAIN", "FRANCE", "ENGLAND", and "ITALY" make translation harder.

Multi-line labels such as "HOLY ROMAN EMPIRE" (Holy Roman Empire (Q12548)), "ATLANTIC OCEAN" (Atlantic Ocean (Q97)), "Mediterranean Sea" (Mediterranean Sea (Q4918)), and "Dutchy of the English King". Looking at SVG source (with lots of Inkscape details) shows that multiline text is done correctly with embedded tspan elements. That may be good text line handling by Inkscape. However, the SVG Translate tool breaks them out as individual tspan elements so the grouping is lost. Better to have "Mediterranean Sea" translated to "Mar Mediterráneo".

Map conventions[edit]

Stolen from Talk:Oroville Dam crisis:

There is convention for maps at Wikipedia:WikiProject Maps/Conventions. This type of map would be a "Technical map" (see that section of the page), for which there is no standard other than the standards for maps generally. File:Maps template-en.svg has fonts and a color palette for use in maps generally (there are specific color palettes for some types of maps on the map conventions page). I used to be active on Wikivoyage and created a few maps, that project's guide recommended DejaVu Sans bold for main labels on maps so I guess that I have continued using that for maps (when I don't use the template svg). There was no particular reason for 11pt. AHeneen (talk) 03:05, 25 February 2017 (UTC)
Thanks for that. Excellent lead- I had never been to Wikivoyage. I loved the bit on Wikipedia:WikiProject Maps/Conventions#Technical maps that said "This style of map doesn't have a convention yet, but needs one." I am now chasing Cynthia Brewer though I may have to clean up that page first. ClemRutter (talk) 11:33, 25 February 2017 (UTC)

Copying is a poor method of using a symbol library. Copies are difficult to maintain. In its defense, SVG xlink: is limited.

Inkscape may get in the way with its styles.

Text selection, space, and hyphen[edit]

Selecting and copying the two-line "Mar Mediterráneo" from the SVG gets "MarMediterráneo"; the copy did not add a space even though the words were on separate lines. SVG should have clipboard copy rules for text and tspan elements. The text element has xml:space="preserve", so that should control. Newlines are supposed to be turned into spaces. However, there is not a newline within or after tspan elements, so no space should be inserted. That would explain why no spaces occur between the separate lines. 

<text
       sodipodi:linespacing="125%"
       id="text4976"
       y="662.791"
       x="126.492"
       style="font-size: 32px; font-style: normal; font-variant: normal; font-weight: bold; font-stretch: normal; text-align: center; line-height: 125%; text-anchor: middle; fill: rgb(100, 100, 100); fill-opacity: 1; stroke: none; display: inline; font-family: Arial;"
       xml:space="preserve"><tspan
         y="662.791"
         x="126.492"
         sodipodi:role="line"
         id="tspan4982">Duchy of</tspan><tspan
         id="tspan4989"
         y="702.791"
         x="126.492"
         sodipodi:role="line">the English</tspan><tspan
         id="tspan4991"
         y="742.791"
         x="126.492"
         sodipodi:role="line">King</tspan></text>

Including a trailing space within the first tspan should insert the space, but it might confuse text-align="middle". A newline might make more sense if it just turns into a space on select and copy: 

 <text xml:space="preserve"><tspan>Dutchy of</tspan>
   <tspan>the English</tspan>
   <tspan>King</tspan></text>

The above is wrong due to "preserve": there would be too many spaces. It controls the initial and final newlines, but the indentation would introduce more spaces. Using xml:space="default" would have the better semantics; it deletes newlines but merges spaces; it also deletes leading and trailing spaces. 

 <text xml:space="default">
   <tspan>Dutchy of</tspan>
   <tspan>the English</tspan>
   <tspan>King</tspan>
   </text>

The above might get the right result; the spaces, however, will come from the indentations. Newlines will be deleted, tabs become spaces, leading and trailing spaces are deleted, and then spaces are coalesced.

It looks like the xml:space issue is more complicated and changing. If I try it in HTML, I cannot produce the expected problem. Many browsers and XML tools ignore xml:space.

Looks like xml:space is deprecated but still alive in SVG 2.

The recommendation is to use the CSS white-space property.

Another issue is whitespace in XSLT.

There's also an issue with hyphens. If a line is broken by hyphenating a word, then it would be nice if the copy removed the word break hyphen but not disturb an ordinary hyphen.

  • The semi-
    conductor diode
  • The Taylor-
    Jones induction coil
  • With styles to suppress selection: The semi-
    conductor diode
    • doesn't work in Edge; hyphen suppressed in Firefox

The checkered career of the soft hyphen:

Argues the semantics started out being the soft hyphen was the thing added at the end of the line by the line breaking agent, but then the semantics shifted to signal a line-breaking opportunity. Points to different standards, offers an interpretation of the standard's meaning, and covers some implementations. For example, &shy; is a real hyphen in Microsoft Office (try cut and paste!); Office uses something else to signal line breaking opportunities; when Office spits out HTML, its optional hyphen is converted to a &shy;.

  • non-breaking (hard) hyphen &#8209 A‑B
  • soft hyphen putting &shy; at end of line does not display hyphen:­
    • A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­A­B­
    • Searches are supposed to ignore &shy;
    • HTML 2 discouraged using &shy;
    • HTML 4 labels it a hyphenation opportunity
    • Unicode no longer labels it as (optional) punctuation
  • CSS hyphens property
    • hyphens: none (do not break)
    • hyphens: manual (at hard or soft hyphens)
    • hyphens: auto (language dependent)
    • https://developer.mozilla.org/en-US/docs/Web/CSS/hyphens
      sez for Swedish hyphenation: Chrome no, Edge ?, Explorer yes, Firefox 43 implementation note 8
      https://support.mozilla.org/en-US/questions/1202353
      Firefox hyphenation dictionaries: resource://gre/hyphenation/
      seems to be TeX hyphenation algorithm which just determines where to split
      https://sv.wiktionary.org/wiki/tuggummi gives hyphenation as tugg-gummi (Swedish hyphenation adds extra g)
      may need lang attribute on html element
      http://caniuse.com/#search=hyphens (April 2017) shows MS Edge wants -ms-hyphens:auto; Chrome only on Android and Mac.
      Try CSS stunt to produce Swedish hyphenation "tugg-gummi":
      G-G: tug­gummi tug­gummi tug­gummi tug­gummi tug­gummi tug­gummi tug­gummi tug­gummi tug­gummi GG- tugg­ummi tugg­ummi tugg­ummi tugg­ummi tugg­ummi tugg­ummi GG-G (sv rule of no consonant triples) tugg­gummi tugg­gummi tugg­gummi tugg­gummi tugg­gummi tugg­gummi GG tuggummi tuggummi tuggummi tuggummi tuggummi tuggummi
  • http://unicode.org/reports/tr14/#SoftHyphen

Text selection:

There's some indication that limited amounts of text may be selected. It may be confined to a text element.

Post translation changes[edit]

Back to the map example.

IETF langtags handed as native language local only; langtags not shown, but it looks like the list is in langtag order.

Resulting labels took effort. Some label layouts have been tweaked.

After diagram was made, somebody added additional cities for time of Joan of Arc. Possible that previous translations did not get the benefit of the additions (see -sv version).

Color scheme is poor, but now exists in several instances. It would be better to choose a good color scheme in one place.

Capitalization[edit]

The map also shows another problem: capitalization. The country is "Spain", but the map text element uses "SPAIN". That complicates string matching. It would be better to use "Spain" and "England" as the actual text (rather than "SPAIN" and "ENGLAND"), set a class "country", and then use CSS style to uppercase the presentation text: 

<style type="text/css" media="media types" title="advisory title" >
  <![CDATA[
    text.country: {text-transform: uppercase; }
    text.ocean:   {text-transform: uppercase;
                    font-style: italic;
                    color: blue; }
    text.sea:     {font-style: italic;
                    color: blue; }
  ]]>
</style>
  • Spain → Spain
  • Atlantic Ocean → Atlantic Ocean
  • Mediterranean Sea → Mediterranean Sea

Interestingly, CSS styles would do their proper role here: they affect the presentation while keeping ordinary spellings ("Spain") would preserve content. The class information could be used as additional context in the translation process.

Possible text-transform values are none, uppercase, lowercase, capitalize, full-width. In addition, a CSS property can be inheirit or initial, but initial is not cap first letter of the string but rather the default (initial) value.

Also, capitalization problems arise elsewhere. For an earthquake diagram, it would be better to use "epicenter" rather than "Epicenter" to aid translation; a capitalize transform would then capitalize the word. It would be sticky for "center of the universe" becoming "Center Of The Universe". It may also have bizarre consequences if other languages do not have same notion of capitalization. It might be better to just avoid the capitalization completely and leave the picture with "epicenter" rather than "Epicenter".

  • eye of the tiger → eye of the tiger
  • epicenter → epicenter

Plurals raise the same problem. The chemical elements include a class of noble gases, and there is some variation in how to refer to the class or membership in the class. A diagram might identify neon as a noble gas; the wikidata class is noble gases (Q19609) (singular). In contrast, wikidata uses the plural for alkaline earth metal (Q19563) with an alias for "alkaline earth metals". Aliases do not cover the missing plural/singular. The issue gets much more complicated for languages that decline adjectives: the string distance increases.

A map spaced letters out: "ATLANTIC OCEAN" became "A T L A N T I C   O C E A N". The spacing can be done with the letter-spacing property. Can apply capitalization and spacing at the same time with style="text-transform:uppercase;letter-spacing:0.5em":

  • Atlantic Ocean → Atlantic Ocean

That opens up a whole area of single-language edits (no translation skills required) to a diagram or map to improve i18n workflow. Editors who are not translators can edit diagrams to be translatable.

Random test of abbrev:

  • TT
  • TT
Multilingual request[edit]
Generates new file[edit]

My understanding is SVG Translate generates a new SVG file from an existing one; it uses neither XLIFF nor switch. It was written some time ago by user:Nikola_Smolenski, but he had moved on ca 2010, so there was a request for a new programmer for continued maintenance.

User:Jarry1250 took over some maintenance; User:Luxo and User:Leyo chimed in. SVG Translate sources are on github, but they haven't been updated in a year and a half.

SVG Translate is clearly a significant effort. In browsing some sources, it has a multilingual user interface. Such an interface would be needed for editors who only speak their language. It also did some rights management. I have not found significant documentation for either the user or the program architecture.

DerivativeFX may be a licensing tool by Luxo. http://toolserver.org/~luxo/derivativeFX/deri1.php

Discussion about DerivativeFX:

SVG Translate apparently does not use DOM but rather pattern matching. Test for SVG file looks at the file extension rather than the file.

When I tried to run it recently, it threw an error at line 395. An array of all languages is being released when that array is protected. Somewhere I had found a bug report.

Wikimedia selection bug[edit]

Looking at the bad case selection bug (eg zh-Hans). The page HTML has 

<form action="/w/index.php">
  Render this image in
    <select id="mw-imglangselector" name="lang">
      <option value="bs">Bosnian (bs)</option>
      <option value="cs">Czech (cs)</option>
      <option value="de">German (de)</option>
      <option value="en" selected="">English (en)</option>
      <option value="fi">Finnish (fi)</option>
      <option value="hr">Croatian (hr)</option>
      <option value="sh">Serbo-Croatian (sh)</option>
      <option value="sr">Serbian (sr)</option>
      <option value="sr-EC">Serbian (Cyrillic script) (sr-EC)</option>
      <option value="sr-EL">Serbian (Latin script) (sr-EL)</option>
      <option value="vi">Vietnamese (vi)</option>
      <option value="zh-Hans">Simplified Chinese (zh-Hans)</option>
      <option value="zh-Hant">Traditional Chinese (zh-Hant)</option>
      <option value="zh-cmn-Hans">zh-cmn-Hans</option>
      <option value="zh-cmn-Hant">zh-cmn-Hant</option>
    </select>.
  <input type="submit" value="Go"/>
  <input type="hidden" value="File:First_Ionization_Energy.svg" name="title"/>
</form>

So clicking the submit button leads to

Apparently the problem is the option value="zh-Hans" should be lowercased when page is generated, or index.php should lowercase the lang parameter.

The problem is more widespread because mixed-case languages are not properly handled:

Does not work (should show Chinese)
Does work (shows Chinese)

TranslateSVG extension[edit]

Different from SVGTranslate. Wants to use switch:

There's a comment about standard form. Could not find it, but there is IsTranslationReady() (switch installed). Recent translation pushes, but not much in development. Find documentation. Does translation control get in the way? Burt; Jarry1250.

Other players. Rilke. Perhelion. Nikerabbit. Sameboat.

Multimedia[edit]

Wishlist... link to other parts

MediaWiki.org[edit]

The apparent goal is this approach is to mark sections of wikitext as translatable while shielding other markup (such as tables and expressions). A <translate> element is wrapped around material that should be translated. In ITS terminology, that element is a span with attribute its:translate="yes". It also has a goal of breaking text to translate into smaller chuncks.

I don't see any notion of category or hint, but it may be there.

I do not see any notion of its:term, but it may be there. I still am not certain whether the publication "/tech/news" should be treated as a name (and not translated) or an idea that should be.

The activity seems to be conservative: a translation and a review. Apparently nothing is changed until the translation is made.

More

JQuery.i18n

Language attribute

Mas[edit]

BIDI[edit]

There are some bidirectional text (BIDI) issues with Unicode that I need to sort out. Unicode has some default processing, and sometimes that default can cause confusion. There are some fences that can be constructed to avoid problems. I need to find examples that display the issues.

unicode-bidi CSS
HTML has dir=ltr,rtl,auto; elements: span, bdi (bidirectional isolation), bdo (bidirectional override)
SVG has direction=ltr,rtl,inherit; elements: tspan,
  • ionization energy (eV)
  • ionization energy (eV)
  • ionization energy (eV)
  • طاقة التأين (eV) — wrong; (eV) should be to the left
  • طاقة التأين (eV) — works; outside span sets dir
  • طاقة التأين (eV) and more stuff — confused; outside span sets dir
  • طاقة التأين (eV) — wrong; (eV) should be to the left
  • طاقة التأين ‏(eV)‏ — wrong; (eV) should be to the left; rlm moves the space left; trying to move span left.
  • طاقة التأين ‏(eV)‏ — works; (eV) in span with dir="ltr"
  • طاقة التأين ‏(eV) — works; dropped rlm after (eV) span.
  • طاقة التأين (eV) — wrong; dropped rlm after space and after (eV) span.
  • طاقة التأين ‏(eV) more stuff — more text adds confusion; dropped rlm after (eV) span.
  • طاقة التأين (eV) more text — works; outer span is rtl; no marks.
  • طاقة التأين ‏(‎eV) — wrong; no dir= in span but use lrm.
  • طاقة التأين (eV) — works; outside span rtl; internal bdi element.
  • cannot see directional marks in pasted text
  • http://www.unicode.org/reports/tr9/
  • http://www.unicode.org/faq/bidi.html
  • http://stackoverflow.com/questions/5801820/how-to-solve-bidi-bracket-issues

Avoid translation[edit]

Consider an SVG diagram that wants to href="some wiki article". The problem is the URI wants to point to the article for the appropriate language: en:Radium or ru:Радий. That can be done with switch statements that give the alternate spelling and the appropriate URI.

If just the chemical symbol is given (Ra), then there is no need to give the alternate spellings (to push the point).

Wiki-independent redirect: https://foo/mumble?wiki=en?article=Radium -> redirects to the AcceptLanguage version of en:Radium.

Radium is radium (Q1128).

  • Statement: Ra
  • Identifier: 7440-14-4 (capital case!) 7440-14-4 (works)
  • Wikipedia: 133 entries; ???; see comments way above about sitelinks

Instead of AcceptLanguage, CSS could grab the lang variable.

How does Serbian work? Are there Latn and Cyrl versions?

rsvg problems[edit]

Bad handling of IETF langtags.

SVG animation[edit]

Translations would also be possible on SVG animations if WP served the SVG files. Animation GIFs have been translated:

Language processing[edit]

Signing[edit]

Phonemes[edit]

Commas and hyphens prevent Edge SSML from working:

Bézier curve

References

  1. ^ http://www.w3.org/TR/SVG2/struct.html#ConditionalProcessingSystemLanguageAttribute §5.8.1
Retrieved from "https://en.wikipedia.org/w/index.php?title=User:Glrx/sandbox&oldid=1137350229"