FED-S camera and lens s/n needed for statistics

My apologies beforehand to all of you who are unfamiliar with statistics (I don't claim to fully understand it either), still I hope I can show you something you haven't seen before!

One of the many mysteries of the FED1 (and probably many other models) is how camera and lens serials could be related and if there is some pattern that would allow predicting a certain lens serial range from only the camera and vice versa.

I have a FED-S 'd' with an odd 4-digit serial and wondered if the lens serial could shed some light on this issue.
See here:
http://ussrphoto.com/Forum/topic.asp?TOPIC_ID=2348

My hypothesis was that the No.2570 might mean that the first and last digit went missing, making the true serial "12570x".

Getting bored while working on my PhD thesis, I instead ran some statistical analyses on the correlation between camera serials and lens serials in FED-S cameras listed in the Wiki.

I got 48 samples - 17x FED-'c', 27xFED-'d' and 4x FED-'e'.

If we plot the camera s/n on the x-axis and the lens s/n on the y-axis we get a scatterplot which looks like this:



While there are some datapoints far out in the middle of nowhere, most seem to group pretty nicely in the middle. The orange point would be my #2570 assuming it is #125700. To avoid making things even more complicated I have not included this camera in further statistics.


I have marked the 3 models (FED-S 'c', 'd' and 'e') in different colors so we can distinguish them. They form nice separate groups.
Now for a simple linear regression fit:



The red regression line tries to fit a linear relationship between lens and camera s/n depicted in the formula below, while rІ measures how well this model fits our actual datapoints - with 1 being a perfect correlation and 0 no correlation at all. We get rІ of 0,78 which is pretty good but not extraordinary. Since many datapoints lie far above or below the line it misses most of them.


Next I did something you should not really do in statistics - I excluded all far off datapoints to see if we could get a better fit and thereby a model that would explain most camera:lens relationships. Of the original 48 camera/lens pairs only 33 remained.
Excluded datapoints are drawn in a darker shade.



The red regression line now crosses most of our datapoints resulting in a rІ = 0,994
An excellent correlation!
The corresponding formula to calculate lens-camera s/n relationship would be:

Camera = (Lens - 15459) x 10
Lens = 15459 + camera / 10

Feel free to try it out on the FED-S cameras listed in the wiki


Still we see that the datapoints form more of a slight arc than a straight line so I fit a logarithmic curve to the same
25 datapoints, which resulted of an even better match of rІ = 0,998



The corresponding formula would be:

Lens = 25200 x log10(camera)-99708

A bit more difficult, but still easy with a calculator. Again you can try it out on any FED-S camera-lens pair in the wiki to see how good the model fits.
The dotted red line is the 95% confidence interval, meaning that according to the model, 95% of samples should lie within these boundaries. Indeed all but one of the datapoints do, only the excluded ones are outside.


Now what does all of this mean?
I'm afraid not very much…

.) I think there is a general relation between lens and camera s/n even though there are exceptions which don't fit in very well

.) The theory that my camera #2570 is missing first and last digits (#125700) seems possible as there are at least 2 examples with similar lens:camera s/n ratios

.) The amount of cameras which seem to have "younger" lenses than average (above red line) seems to be about the same as the amount of cameras which seem to have "older" lenses than average (below red line)

.) By no means do I claim that this model is suitable to determine "matching" or "wrong" camera-lens pairs, we know virtually nothing about how they were paired in the factory. All we can say is, that at least the 2 points high above (1c #76862 / #27890 and 1d #107723 / #31185 as well as the point far below the regression line (1d #158978 / #24761) seem to be quite unusual.

.) There might be a nonlinear relation between camera and lens s/n indicating that in the beginning more lenses than cameras were produced but this is more than speculative.

.) Finally: all of these results are likely to change with more datapoints.
The 48 camera-lens pairs are a good start but twice as much would be a lot better. If you have a FED-S or know someone who does - please submit the camera and lens s/n to the wiki.
If you don't want to reveal the full s/n for whatever reason you can replace the last digits with "x", this is irrelevant for statistics on this scale.

I'd love to do the same thing for "normal" FED1 but therefore I need you to provide a lot more camera:lens s/n to the wiki, currently there almost none listed.

Regards,
Christian

[EDIT]
I managed to restore the images but only from May 23rd 2013 onwards. Therefore the graphs above already contain seven cam/lens pairs which were added by the posters below.

Hi Christian!

All my thanks for your work. No doubt it will be useful for all the collectors of Fed 1...

I had tried to do something like your work some years ago. My aim was to see if there is a regularity of distribution lens/body. At that time, my guess was that Fed (or another factory) had made the c. 13000 2/5cm lenses in two batches of c. 5000. And a last one, a bit later, of c. 3000.

Anyway, there is something strange. The last 3000 lenses are in the range 30/33000. And this range is occupied too by the 6,3/100mm (from 30xxx to 39xxx). Two lenses at the same time in the same range: it's impossible in a planified economy. Hence the supposition that they had to make more S than planified. Why?...

As for S-s, absolutely no regularity. We meet them from the early 1c to the last 1e. And I was told that some rare early 1f have the 1/1000th too. Exactly like the early 1948 Fed Zorki: KMZ won't use the 1/1000th on their Zorki 1 later... The 1/1000th shutter boxes used by KMZ have probably a prewar Fed origin.

I had tried too to establish a sort of "truth line" which roughly corresponds to your first graphic. But I know nothing about statistics...

Good luck for your work, and thanks!
I will look at the wiki to see if I can complete the Fed S section.

Amitiйs. Jacques.

Thanks for your comment, Jacques.

I doubt my statistics are of much use at the current level but maybe it may lead us to something we haven't seen or thought of before.

Can you give us some more details about your theories?
Why the 2 batches of 5.000 lenses and one of 3.000?

Also you are right, 50/2 and 100/6.3 serials do overlap.
I didn't notice that before and plan to create a graph on all these lens serials in the next days.

The most straightforward explanation which comes to my mind is that the demand for the 50/2 lens was greater than expected when production started so they ended up with overlapping the serial range of the 100/6.3.
Remember that the statistics suggest that in the beginning lens s/n increased faster than camera s/n.

Looking at the distribution of the lens numbers we have (I updated the wiki on the 50/2 lens), it seems all serials between 20.000 and 33.000 were used and there were no major gaps.

This would mean that about 13.000 50/2 lenses were produced but (at least that's what is said) only about 2.000 FED-S cameras. Personally I think this number is too low, but let's assume it to be true for the moment.

This leads to the main question:
What happened to the 10.000+ surplus lenses?

Of course they could be used on any FED camera if collimated correctly but I've never seen a "standard" FED1 with the 50/2 lens.
However they are being sold separately now and then, I found 3 lenses on ebay. I'm not sure if this is enough to account for the "missing" lenses, but it seems there is about the same amount left as of the 28mm, 50 macro and 100mm lenses, which is not really much considering that about twice as much 50/2 lenses should have been produced.


As for the FED-S serial numbers - also no clue.
It seems they were producing one now and then when they were in the mood but without any pattern.

.)Maybe the mechanism for the 1/1000 was difficult to produce or required special materials (I have no idea about camera machanics ) so they only were able to make small quantities now and then.

.) Or they were just produced on demand for special persons/occasions.

Perhaps we'll never know, I wish there was still someone who worked in the factory left ti interview.

Regards,
Christian

I think that 10000 Fed 2/50mm were planned first.
As it is the only lens put on S-s, that means that 10000 S-s were to be made first. And probably due to the demand, some 3000 more were delivered (with 3000 2/50mm more made, of course).

So, some 12/13000 S-s (from 1c to 1e) and 2/50mm were made. About the 1/1000th shutter box, it is interesting to note that KMZ used this speed on some on their 1948 (and perhaps 1949) Fed-Zorki. These shutter boxes probably come from Fed S surplus. KMZ didn't use that speed after on the Zorki 1 and the paint of their shutter box is bright, not crisp like on 1/500th KMZ shutter boxes.

Of course, no proof about all that. But it would explain why 2/50mm are very hard to find alone. For my part, I never saw a 2/50 for sale alone with a special passport...

We had several discussions about this question some years ago. But the wiki was far less important...

Amitiйs. Jacques.

Hi Christian

Here are more Fed-S details from my own collection

Body Fed-S (type D) no 131794
Lens Fed 2/50 no 28904

This seems to fit well with your other plots. Very interesting thread - thanks.

ATB Chris.

(PS - I've added this camera to the Wiki entry, but the table seems to be operating in double line spacing and I can find no way to get rid of the blank line. Can someone help please ... ... Thanks)

Christian,

I have just added some Fed S I had still in my data. Most part come from eBay.
I had tried to find a "line of truth" between Fed S bodies and 2/50mm lens. A sort of regular arithmetic progression corresponding to your red line. 15/36 (in my data) were OK.
But you do far better with your "confidence interval"!

Jacques.

Chris,
thanks for your contribution!

The only way to prevent the double line spacing when editing the wiki is to click the "Source" button, this way you see the entry in code. Probably there is a better solution, but I don't know much about website design.
I'd love to have the option to put the serial data in tables so they are easier to edit and more comfortable to view.

----------

Jacques,
so you think about 12.000 FED-S cameras were made?
This would make sense but why is everyone stating that not more than 2.000 were produced?
This is found on almost all related websites including Wikipedia. Is there a source for this statement or how was this conclusion reached?

I've read that it was indeed very difficult to produce the 1/1000s shutter setting as the gap between curtains is extremely narrow at this speed which makes it difficult to achieve an even exposure.
This is probably the reason why the design was given up soon after the war.

Another interesting question came to my mind recently:
How was quality control handled at the FED factory?
I assume it was still quite good in these days, so what happened to lens/cameras that did not pass the test (an the corresponding serials)?
As these faulty items could not be sold they were probably either repaired or, if impossible, disassembled for spare parts, which means that not all serials really correspond to equipment which left the factory.
Unfortunately I have no idea how much percent are typically discarded in quality control but probably in the beginning the amount was quite high and decreased as workers got more experienced.

----------

Also, Jacques,
I noticed there are some FED-S with Sonnar lenses (also seen in FED-NKAP).
http://www.dvdtechcameras.com/collect/fed/1/5.htm

Interestingly one FED-e in the wiki has a Sonnar-serial which seems to overlap with those of the FED 50/2 lenses.
1e (BERDSK)181306 33048 Sonnar Cl. Asquini.
I don't know if this is a mistake but I'll exclude this entry from my statistics.

I don't remember if I saw this number of production (c. 12000)somewhere else. But for me, it is the only way to calculate the number of S-s. One must remember too that it's almost impossible to find a 2/50 lens or a S body alone...

About the 1/1000th speed, you are right to say that the gap is difficult to regulate. Fed's tools were less precise than Leitz's ones. For example, Fed used till the end bolts with two notches only to regulate the shutters (4 for the Zorki 1, 4 then 6 for the Leica 2), S-s included.

I think too that the S is probably the first camera to be standardized (lenses are dedicated on plain NKVD). The 4,5/28mm and the 6,3/100mm were specially made for the S and it would not make sense if lenses were not really interchangeable. Leitz had done it as soon as 1932 (we are in 1938) and in that matter, it's too a question of precision of machine tools.
I made experiments on a S body which works well (different "S" lenses mounted on this body) and the pictures are OK. But of course it is not enough...

As for Claudio Asquini's 1e S, the lens is a Sonnar block mounted on the rear part of a 2/50mm Fed. The serial number belongs to the Fed ring. It's me who put it in the wiki.

Amitiйs. Jacques.

Thanks for lifting the mystery of the FED Sonnar!
This means an original lens with this serial existed, I will add it to my statistics.

I do not think the FED-S cameras were really standardized like Leicas.
Maybe camera and lenses could be matched at the shop where you bought the additional lenses, they probably also repaired cameras so they should have the knowledge and equipment.
Don't have the means to measure the lens register of my cameras, I'll see if I can get a caliper from university. My 28mm f/4.5 lens would not focus at infinity on 2 cameras. I had it (as well as the 100mm f/6.3 and the 50mm macro) collimated for the camera I use for fotography, now they work perfectly. Yuri from Fedka told me it was in fact easier to re-collimate lenses than to set the cameras to the standard register of 28.8mm which I originally wanted him to do.

Hi! My Fed-S attestat:

camera s/n: 65839
fed 50/2 lens s/n: 21453

Many thanks!
An attestat (or a passport) for a Fed S is always precious and rare!

And welcome here...

Jacques.

Very nice, Serhio!
Thanks for your contributions to the wiki.

The most interesting recent find are 2 very close type 'd' serials, the points almost completely overlap.:

Camera / Lens
131794 / 28904
132262 / 28946

I hope we can get some more FED-S type 'e' serials, we only have 4 until now.

Regards,
Christian

quote:

---

Originally posted by vorobaz

Hi! My Fed-S attestat:

camera s/n: 65839
fed 50/2 lens s/n: 21453

---

I have Fed-S camera # 65769 and f50/2 lens #21118

I should add:
Fed-S camera #145843; f50/2 lens #24466

I also have an unusual body (#172220) that has the 1/1000th speed marked with a line but the speed is not engraved. It came with an f50/2 lens (#28534) that is coated and is not in register with the body but works well on post-war Soviet LTM bodies.

Thank you for your great charts

Very interesting!
Really, we lack these Fed S passports.

About your s/n 172220, does the 1/1000th work, even if the speed is not engraved? I suppose it is always a "d"? The "e" begin at c. 173600.

Thanks! Jacques.

Thank you, Michael!

I will integrate your data in the next update of my graphs, we already have a couple of new camera-lens serial pairs.

In both of your cameras, the lens serial is unusually low compared to what we would expect from the camera serial - or the other way round.

Could you provide pictures of the unusual camera and lens?
The #28534 lens serial lies amidst the usual range of FED "d" cameras, so it seems strange that it is coated and registered for post-war cameras.
Your camera #172220 would be the highest FED-S type"d" serial we have in the wiki.

Regards,
Christian

Another strange FED-S
http://www.ebay.com/itm/The-rarest-russian-camera-FED-S-NKVD-1-1000-lens-2-50-Leica-copy-case-2092-/181183459655

Type'c' No#2092 with lens serial #33128

This is the 4th FED-S with serial engraving error, also the lens would better fit in the late Berdsk-range than for Type'c'.

I am not convinced by the "engraving errors": they are too numerous if we add the chapter "Fed 1b-unusual serial numbers" in the wiki.

We had discussed about that question a while ago. If I remember correctly, we had found two possible explanations: important repairs under warranty or upgrading at any time. Leitz had done just the same for their LTM cameras: put the former serial number on the new camera.

That would not exclude the always possible engraving errors, of course...

Jacques.

Indeed, we simply do not know what happened here.
The camera #2092 seems to be a mixture of different parts.

The problem with the repair / upgrade scenario:
Why do we mostly see 4-digit serials?
Why no FED'e' with late 'c' or 'd' serial or vice versa?

The war-time commissar's camera I inherited from my grandfather might be such a case.
It has a FED-1'c' YCCP top plate #60570 / Lens# 99697/55
and a brass shutter cage with a hole in the pressure plate.
But all other features are from FED-'d'
.) Big bolt for bottom plate holder
.) No screws under rewind knob
.) top front center screw (near the lens rim) completely visible

He had the camera repaired in Vienna after the war, as the shutter was not working properly, so some parts may be from Leica. He guessed it to be from approx. 1936 which would even be too early for the serial.
I'll post photos in a separate thread.


As for My FED-S #2570 / Lens# 25237/10:
I've sent it to Yuri for CLA and asked them to check for unusual parts - they found none. It's a normal FED'd'
.) Shutter cage: aluminium
.) Number '5' stamped on bottom of shutter cage
.) Hole in pressure plate
.) Rivets in catch of the lock for base plate
.) Big bolt for bottom plate holder
.) No screws under rewind knob
.) top front center screw (near the lens rim) completely visible

I have no idea what to make of this.

I have only some general ideas to propose.
Of course, all we can say about all that is "perhaps" or "maybe"...

1- Upgradings are possible. Zeiss and Leitz did so for their Contaxes and Leicas I and II. So, why not Fed? I have, let us say, a 1a. I want an S and give my 1a in exchange (and money!)

2- Repairings are possible too. Zeiss repaired complete series of their Contax I which was absolutely not reliable. We can imagine the same for Fed which had not experience in cameras. So, I resend my 1a which does not work and I receive a perfect 1b with the ancient s/n. Under warranty. But why not after warranty, too?

3- Errors too are possible. For sure, we know some: a "5" more stamped on a lens, a cipher more on an accessory, two letters less on the plate of a Fed 2... But certainly these errors are not so numerous: difficult in a planned economy...

4- Of course, we were not there on those days. But nowadays the rare Fed are sought after and it is not very difficult to make one from spare parts and sell it at a good price... BTW, the 1a we spoke of in another thread has disappeared: sold!

Amitiйs. Jacques.

Time for an update...

We now have 54 FED-S lens/camera pairs:
18x FED-S'c'
30x FED-S'd'
06x FED-S'e'




We still see that many datapoints are clustered along a line in the middle, but instead of a linear and logarithmic relation between camera and lens serial number it looks more like we have two subsets. One for 'c' (pink) and one for 'd' (blue). Again 'outliers' were excluded for the regression line.






The relation between FED-S'c' serials seems to rise steeper than the relation between FED-S'd' serials.




Not much can be said for FED-S'e', still only 6 widely distributed datapoints. It seems a wide variety of earlier lenses were used for the FED-S'e'


Now what can we do with these datasets:
The steeply rising red line was laid through the FED-S'c' datapoints clustered in the middle the ones far out were omitted. If we believe it to represent the "usual" relation between lens and camera s/n, and assume that the lowest lens serial was #20.000 we can extrapolate that the lowest FED-S serial should be in the region where the red line cuts the x-axis, around #53.000. This matches nicely with the FED'c' serial range starting around 54.400 according to sovietcams.
This serial range also matches with the start of FED-S production in 1938.

The lowest FED-S camera/lens pair we have is around 59.000 but according to the wiki we have several cameras in the range 22.000 - 25.000 and one around 40.000, unfortunately without lens serials. Do we know if any are true 'b' instead of 'c'?
The #25949 is a FED'c' for sure.

So there are more serial mysteries, apart from the 4-digit ones...