Cheers, Jacques!
'Impossible' is such a harsh word, let's call it 'difficult' instead!
First of all my apologies, I can't believe so many years passed since I last visited this forum.
I wanted to restore the graphs long ago but could not find them anymore and then outright forgot.
Been busy with a million other things in the past years and just today thought about checking on the state of FED1 research.
So while I cannot restore the old images - it makes no sense anyway as they are completely outdated - I can make new ones with the current state of wiki entries. As filehosts on the web don't live forever please save these graphs so you can repost them, maybe put them on the wiki. I made them for you, they are free for all.
Omitting all strange serials, we currently have 97 camera-lens pairs for the FED-S with FED-50/2 lens. 33x FED'c', 56x FED'd' and still only 8x FED'e'.
I cannot even remember the old graphs but I believe we see more or less the same here.
Camera-lens pairs concentrate along a vaguely logarithmic line in the center with outliers above for early cameras and more outliers below for middle and later ones. So far not very surprising.
Now let's divide the datapoints into models 'c' 'd' and 'e'.
Full dots are what I arbitrarily considered 'outliers'.
Immediately we see the 3 generations form distinct clusters.
FED'c' (red)
The earliest 'c' cameras make a nice cluster, a few late ones were fitted with lenses of very high s/n. Only two late cameras have unusually low serial lenses. Late 'c' cameras don't group well, there's quite an gap between #80.000 and #95.000
FED'd' (green)
Unsurprisingly the largest batch.
Interestingly from earliest to latest 'c' numbers we see camera-lens pairs clustering along a narrow perfectly strange line. There is no way this is a coincidence, I think we can safely assume these pairs were sold together. We see some lenses with unusually low s/n for early 'c' both high and low for the middle and as one would expect quite some lower s/n lenses for late models.
FED'e' (blue)
Unfortunately we only have eight camera-lens pairs for this enigmatic series, one additional lens has only 4 digits so I omitted it. And we have no lenses for the latest known 'e' models at all.
We have five pairs clustering along the same line as the FED'c', three of which are extremely close together. I would also see these as factory-matched but with so little samples it's hard to tell. Three more have far lower serial lenses, two of which are extremely low and more in the expected FED'c' range.
Now finally I again did some bad statistics and excluded all outlier-datapoints not clustering together and made a separate regression line for 'c', 'd' and 'e' cameras.
What we see immediately is that camera-lens pairs behave differently for 'c' and 'd' models.
The best fit through the 'c' cluster is much steeper but the cluster is fuzzier than the almost perfect straight line in 'd'.
What does this mean? It looks like when lens production started a lot of FED-S'c' were produced hence the gap between camera serials was smaller than in later 'd' models. I'll check this with normal FED1 against FED-S serial numbers. Also it looks like in the lens range #24.000 - 25.500 they had not enough cameras and these lenses ended up in later 'd' and even 'e' models?
Whatever happened, upon production of the FED'd' some different and fixed ratio between camera and lens serial was applied and strictly followed at least for the majority of cases, possibly until at least mid 'e'. The 'd' and 'e' regression lines are basically identical but the data for 'e' cameras is very poor. I did exclude the 'e' with the highest known lens serial number above 33500 as it was only listed in the lens wiki. It falls well within the 'd' line but would mess up the data for 'e' because we only have 5 samples.
Also note that pairs off the regression line never cluster together, they are more or less randomly distributed and far apart. In my opinion this indicates these pairs were not formed in a factory in an organized manner but probably much later as repair or upgrade with random available parts at different places. Maybe even cobbled together recently from parts for sale. We don't know what happened but it’s relatively safe to say it was not an intentional and organized matching of lens and camera serial numbers in the cases of these outliers.
Still keep in mind we have only 97 samples which, given an assumed production number of 13.000 FED-S and 50mm f/2 lenses is about 0,75%. Not great, not terrible. The clusters are highly significant in any case.
So, what do you think?
