IBM "Shop Date"

[phosphorglow]

This has confused me for a while. And apparently many others as well. Some people say there's a week of production code in there, or it's a variation of a Julian date. So I started a spreadsheet of Shop Dates.

And then I just stumbled over this photo of an internal label on a Model F where a worker wrote the date next to the "Shop Date"

2014-03-18 22.21.49 by Sean Forrester, on Flickr

5/29/84 = 4829

So I started adding and subtracting in my spreadsheet and had a theory: what if it's based on working production days?

There were 252 working days in 1984, with holidays accounted for.

So, let's jump to 1989. An example Model M with an internal shop date of 6096 and a case date of 4/03/89. The correlation between the two is approximate, because the sub-assemblies weren't manufactured the same day the label was slapped on the case.

So! 6096-4829=1267

1,267 working production days between them.

How many working days between 1984 and 1989?
'84 - 252
'85 - 252
'86 - 251
'87 - 251
'88 - 251
1,257

10 days difference can't be coincidental.

So!

Let's go back in time, where does shop date 0001 bring us? Well, I didn't get exact with working days through the years and used an average of 250, bringing us back to 1964.

"In 1957, the Electric Typewriter division also completed transfer of its manufacturing and engineering operations to a new plant at Lexington, Kentucky.

To indicate more accurately the scope of the product line, the Electric Typewriter Division changed its name to Office Products Division in August, 1964"

https://www-03.ibm.com/ibm/history/exhi ... ffice.html

It's the best theory I can come up with, anyways.

[phosphorglow]

Adding a little more...:

Late summer 1990 marked the end of Lexington's ties to IBM's manufacturing operations. Lexmark took over IBM's operations in Lexington on 21 March 1991.

https://books.google.com/books?id=wiVmR ... rk&f=false

I have a lot of pictures to plug into my spreadsheet, but I can say that the early style internal label with the shop date appeared on a board from 1/23/90, but not on a board from 6/11/90, which falls in line with the transition period mentioned above.

[clickykeyboards]

Good theory. Needs more data. IBM was very particular about assigning unique serial numbers to model M keyboards. I always thought it a sign of pride in craftsmanship that model M keyboards have hand-written initials of the those who assembled the masterpiece.

I would like to see more data though on the "shop date" theory. I can send some additional photos later of the various 1985 and 1986 model M keyboards that I have later next week.

When I first had the idea of focusing on model M keyboards and ClickyKeyboards in 1998 and then formally started collecting data in the 1391401 database in 2004, I focused on the data on the exterior rear keyboard label. The idea was that I could detect a pattern in the serial numbering and ultimately get an idea of how many model M keyboards were produced.

After 12+ years of data collection and 2000+ entries and having spoken with even a few grad students, post-docs and tenured faculty in applied math and comp sci, they have plotted and looked at the collected data and found some basic observations. No duplicate serial numbers were found. Model M keyboards seemed to never have been made on Sundays or on US holidays, rarely on Saturdays. However, the variability in the numbers when plotted doesn't lead to any conclusions on average production/year or total production.

raw data for 1391401 keyboards
https://docs.google.com/spreadsheets/u/ ... 6ZYHwI/pub

[jeb]

Model M keyboards seemed to never have been made on Sundays or on US holidays, rarely on Saturdays.

Are you referring to the date on the back label? Because your spreadsheet does contain some examples with a date that was a Sunday; I analyzed this at one point.

[clickykeyboards]

@jeb, Good catch. It's been a couple of years since I had someone look at the data carefully and the users are constantly adding their model M data.

For the 1391401, IBM used a few different date formats

For example
02MAY89 http://clickykeyboards.com/product/1989 ... ong-cable/
24JUN91 http://clickykeyboards.com/product/1991 ... -jun-91-3/
05-22-92 http://clickykeyboards.com/product/1992 ... -05-22-92/
02-11-92 http://clickykeyboards.com/product/1992 ... m-2-11-92/

..and sometimes for the last case, the information gets mis-typed from February 11, 1992 to November 2, 1992.

But thanks for looking at the raw data.. glad to know that it helped someone out after running for the past 12 years.

[phosphorglow]

@clickykeyboards:

Definitely agree about the internal labels and the initials of the workers. It's a wonderful thing!

More data, 'eh?

Here's a fun spreadsheet that's editable by anyone, because that seems like a good idea. Add in a shop date and see if it falls close to your case label date.
https://docs.google.com/spreadsheets/d/ ... li=1#gid=0

My math is a little silly, but I think it's the best I can get based off of a single date and not knowing the exact days IBM counted as production days. But it's pretty freakin' close. I have plenty more I need to add to that list, but 20 is a good starting point.

[bitslasher]

This is good stuff guys! I love it when folks dig in and start peeling back the layers of the onion. It would be really cool to know what all those codes mean. You'd think that the Unicomp folks would have some knowledge of this.

[Panter]

I'll definitely be keeping an eye on this topic! This is the kind of stuff you don't read about on most other wiki's (and it probably will take a while, and a lot more research, before it will be added to the deskthority wiki:P).
Anyway, about the initials, have any of the names ever been tracked down?:)

[pr0ximity]

Can't quite figure out how you're organizing that spreadsheet, but here are two interesting examples:

My 19th of May 1986 1390120:

Your spreadsheet seems to put it around late May '86 or so (1.98 years from June '84)

My dateless 107-key 4704 model F:

Spreadsheet puts it around mid December of '86 (2.55 years from June '84) so a bit younger than the Model M!
Also fun to note: the "Final Insp." has the same initials as I do

[clickykeyboards]

1986 IBM model M (1390131)
exterior label: 26APR86
interior label shop date: 5330

[snuci]

Here's a few of mine:

Date: 13JAN86
Shop Date: 5257

13JAN86-5257.jpg (213.4 KiB) Viewed 9746 times

Date: 13JAN86
Note: A serial number?

13JAN86-2.jpg (138.8 KiB) Viewed 9746 times

Date: 19FEB86
Shop Date:5283

19FEB86-5283.jpg (232.17 KiB) Viewed 9743 times

Date: 09MAY86
Shop Date: 5338

09MAY86-5338.jpg (217.63 KiB) Viewed 9746 times

Date: 22APR89
Shop Date: 6110

22APR89-6110.jpg (122.25 KiB) Viewed 9746 times

Date: 03FEB90
Shop Date: 03 Feb 1990

03FEB90-03FEB90.jpg (182.46 KiB) Viewed 9746 times

[terrycherry]

[US][AT][1986.03.25]IBM Model M 1390120(Right corner grey metal logo white case)

[clickykeyboards]

I locked myself in my workshop this weekend and I had a good 8-hours of productive model M restoration work.

Including a date-set of 1986, 1987, 1988 and 1989 model M keyboards

Here is some additional primary research data. Hope the additional data is useful.

1390120 27 MAY 1986 back label

1390120-27MAY86-BACK.jpg (455.71 KiB) Viewed 9675 times

1390120 27 MAY 1986 internal label

1390120-27MAY86-INSIDE.jpg (420.86 KiB) Viewed 9675 times

1391401 30 OCT 1987 back label

1391401-30OCT87-BACK.jpg (399.06 KiB) Viewed 9675 times

1391401 30 OCT 1987 internal label

1391401-30OCT87-INTERNAL.jpg (351.45 KiB) Viewed 9675 times

1391401 06 JAN 1988 back label

1391401-06JAN88-BACK.jpg (374.05 KiB) Viewed 9675 times

1391401 06 JAN 1988 internal label

1391401-06JAN88-INTERNAL.jpg (376.06 KiB) Viewed 9675 times

1391401 30 MAY 1989 back label

1391401-30MAY89-BACK.jpg (450.49 KiB) Viewed 9675 times

1391401 30 MAY 1989 internal label

1391401-30MAY89-INTERNAL.jpg (449.98 KiB) Viewed 9675 times

[phosphorglow]

Ooooh! Goodies! I had to walk away from this spreadsheet for a bit.

Thanks for the info everybody!!! I'll plug 'em into the spreadsheet and report back with my findings.

[phosphorglow]

Alright! If we think of the spreadsheet as a big dart board, and your shop date as a dart, you're guaranteed to hit within a two month area in relation to the case date.

The formula brought three of them super close, basically to the same week.
Otherwise, there's about a month variation, especially the earlier you go.

(Another fun thing to loosely factor in might be the ink stamps on the inside of the cases. My September '85 M has stamps from mid July.)

Thanks for posting pictures of the labels, by the way. I love seeing all of them in one spot like this!

[phosphorglow]

Can't quite figure out how you're organizing that spreadsheet, but here are two interesting examples:

My 19th of May 1986 1390120:

Spoiler:

Your spreadsheet seems to put it around late May '86 or so (1.98 years from June '84)

My dateless 107-key 4704 model F:

Spoiler:

Spreadsheet puts it around mid December of '86 (2.55 years from June '84) so a bit younger than the Model M!
Also fun to note: the "Final Insp." has the same initials as I do

It's organized in roughly the same way I think - all over the place.

At first I had to reacquaint myself with it and think for a second. Month counting in the formula goes like this:
June to July = 1, July to August = 2, August to September = 3, etc...

So!
5/19/86 : 5346 : First week of April '86.
?????????? : 5492 : Third week of November '86.

(Also: apparently for the spreadsheet to make sense, you need to listen to 'The Bangles - Walk Like an Egyptian' on repeat. Bonus points if you dance to it as well.)


It's silly and confusing, but it somehow works to a relative degree. Or maybe it just works! Maybe they did manufacture more sub-assemblies than they stuffed into cases for the month? Questions to keep us awake at night, these are.

[dad]

My shop date is K2 5085 (Model F). BTW it is the same as the last 4 digits of the serial number...So nobody knows the exact date, I guess it is somewhere close to May 7 1985.

model_f_label.jpg (1.04 MiB) Viewed 8453 times