Recently FifthAvenueRestorations (Mark) posted his observations on case serial numbers verses movement date codes. His observation was that the first number of the serial number located on the inside or outside of the caseback more often than not confirmed the movement date code on earlier Bulovas.
I took it one step further and started comparing the watches I have entered in the known and unknown database, and found a very consistent correlation on my watches dating up to 1946. It starts to get a bit foggy between 1946-1949, when some watches were stamped on the back with the letter/number date code and others were not. From 1950 on, its a no brainer.
The purpose of this post is to get the thoughts & opinions of Admin., Panel and non-panel members on the merits of case serial numbers, and if we should change the way we date our pre-1950 watches.
Our current policy is to date with the movement, but some members have been using the case serial number when it did not agree with the movement date code. Should we change this policy??
I am inviting both panel and non-panel members to discuss this further and decide on a uniform way that all Bulovas will be dated instead of this way and that way. Maybe after a week of discussion, we should have a panel member and poll vote on this and let the outcome determine the policy.
Mark's original post http://www.mybulova.com/node/3495
Greg
My feelings on the matter is I believe Mark has a point, but that's it and more time will tell. Is it definitive, NO, it's a tool, and like any tool, it's only as good as the person using it. I believe we have "FEW" tools at our disposal and we should use them all carefully. Just remember that naming a watch based solely on a case number and not in conjunction with movement and date code is a fools errand especially with the ability of a movement swap.
Thanks Greg for looking at that pattern. With all the watches in the database, one could have a large enlough sample size to make some pretty strong statements about these patterns. I too when through the "hard way" and looked at 50 cases/movements, but it was like pulling teeth to go in to each record and extact each field, etc... and I just deferred to what the experts here already know from experience.
If these data were made available in a less "clunky" format, I would volunteer to help run some tests on them. I've got all the various enterprise software with which to do the tests. I understand the assumptions which need to be met so that we arn't "armwaving". It's really pretty simple, and may remove some of the "doubts" with results. This is one of the things I do for my day job, so am comfortable with this stuff.
However it may be better if we request admin, although he's got his hands full, to take a look at this? I can't provide the software to others, but his existing DB software could do some discriptive stats and probably give a coorelation coefficient for some sets of the data.
In reply to Thanks Greg for looking at by William Smith
Will,
I did a little of this the hard way, I just used the known and unknown database and searched watches by year, It didn't do it to all years, only about 6 or 7, I just did it random...I found that on a majority, no case serial is listed, on the one's that had serials, 7 out of every 10 matched the date on the movement, and those that did not, were within 2 years - and 1 year + of the case serial, a couple had weird serials like 123456, and that told me that someone just was to lazy to ad it, so I ignored it.
It would be interesting to do it to the whole database, I bet the results are about the same.
Well, I see more that just a concidence on the 70 that I have in the database, out of that 70, 31 are per '46, and out of those 31, 5 have a serial number that does not agree with the case number, and out of those 5, the movement year is 1 year + or - of the serial number except for the Curtis I just entered, that has a '27 movement, and a '30 serial...Movement swap is my thought on that one.
I also think it would be acceptable if the movement was 1 year newer than the case, example...in 1940 they made 10,000 Bruce cases, and 30,000 10AX 15J movements, which also went into the Bankers, Aldens, plus others... maybe the Bruces were not selling so good and the Bankers were selling like hotcakes, guess what, at the end of the movement run, they had 1,000 1940 Bruce cases left, scrap them, no, just put '41 run movements in them and sell them as '41 models. Vice-versa for the same scenario with leftover movements, just stick them in the new model year cases.
In reply to Well, I see more that just a by OldTicker
I think the issue of whether the serial number is a valid and reliable method of dating a case prior to 1946 is important and needs to be settled. After we settle that, though, I think we have some specific scenarios to address, such as:
Pre-1947
- When a movement predates the case serial number, which rules?
- When a movement post-dates the case serial number, which rules?
- If there is a difference in case serial and movement date of more than two years in either direction, does that force one or the other date to rule, and, if, so, which one? Should this determination be based also on whether it is a known model with a known date range?
Post-1946
- When the movement predates the case date code, which rules?
- When the movement post-dates the case date code, which rules?
- If there is a difference in case date and movement date of more than two years in either direction, does that force one or the other date to rule, and, if, so, which one? Should this determination be based also on whether it is a known model with a known date range?
There may be other scenarios that need to be addressed. At this point, we're just throwing out thoughts for further discussion.
Doing some descriptive stats on these data would be very simple. About the most simple "test" on could do. Nothing fancy.
Admin may take a look for us, if it's waranted. This would simply replace our beliefs with results.
I'm not talking about rules for dating models. i'm just thinking about relationships.
I am interested in the serial date code theory and accept that it should be discussed. Dating watches using a theory however, is a disservice to the users of this database. I cannot, at this time ascribe to this theory. My suggestion:
1. No date case but known model - we go by the mvnt UNLESS it does not fit within the years the known case was produced, in which case, the year of the watch - forget about the mvnt code - will be the last year the case is know to have been made - eg, cut corner with '32 mvnt is a '28 LE
2. No date case but known model with a mvnt that fits the years known to have been made - mvnt code is date of watch - eg cut corner with a '28 mvnt is a 28 LE
3. Dated Case with a dated mvnt - latest determines year unless mvnt does not fit within known record. eg - AA case (the ribbed model date L1) with a 1960 mvnt is a 1951 AA
4. Dated Case with a dated mvnt that is within known production years - the latest date - eg - AA case (the ribbed model date L1) with a 1950 mvnt is a 1951
The 1-2 year swing mentioned in several of the preceding post is not troubling to me in the least as we regularly see the same deviance in dated case/mvnt watches.
William, in my opinion, the issue is not that simple, as there are other questions that need to be addressed. As OT implied by the title of this post and further explained in his opening statement, we are also concerned with what happens when the case (however it is dated) does not match the movement. If they match, then it's a "no brainer".
We have a number of different scenarios that we are presented with on a daily basis, and we need to address those so that we have a consistent method of dating a watch that presents with any one of those scenarios.
Database statistics may help folks feel comfortable (or not) with relying on the serial number as the date for a pre-1946 case, but that is just one piece of the puzzle.
The theory is backed up by a trend, on which I form My opinion. Most Watches I own fit the 'no brainer' category.
For those that don't:
If the Movement post dates the Case then the Movement prevails.
If the Case post dates the Movement then the Case prevails.
within reason.
Now You have this 'tool'.
If the Case doesn't match the Movement don't purchase the Watch.
In reply to The theory is backed up by a by FifthAvenueRes…
As for your "tool", that assumes we always have access to that information, which we quite often do not. Most sellers don't know that the movement has a date symbol, much less how to interpret it. I have found that, even when I ask them to tell me the symbol--and even give them examples of what it might look like--the response is that no such symbol exists, which, of course, turns out not to be true. The rest of the time, I get no answer at all. It's a crap shoot when you buy a watch sight unseen, and we need to be prepared to deal with the consequences.
The images are very rarely clear enough to see the date symbol, as I'm sure you know.
If its a model that I haven't see before, or very rarely, and all other features look genuine, then I'll take the chance. Every time.
But I think this discussion is getting a bit off track. . . .
The purpose of this thread is to find a uniform way for members to date their watch, not for one to do it one way and someone else to do it another, all watches should either be dated by the movement in the case, or case serial number, not both unless they both are the same year.
Then that would be the latter Date IMO Greg, within reason.
We are already doing so in many instances:
eg - 'ACADEMY AWARD' series Watches with movements Dated A9 and Cases Dated L0 are Dated 1950 in the database.
If one were to have an L0 Case and a Movement Stamped L1 We would assume 1951.
Correct, if the caseback is stamped, that is what we go by, but the real reason for this post is to get the comments and thoughts of others on whether or not the pre 1950 watches without a date stamp on the case should be dated by the movement or serial...I think the serial has merit, and what we need to decide is do we change the way we date the pre date stamped watches and make it a uniform rule, or just let everyone do what they wish???
I personally don't think the latter makes for a reliable database...
In reply to Correct, if the caseback is by OldTicker
Yea Lisa, it is but one piece of the puzzle, but my suggestion not only gives us statistics, it would allow someone (maybe me- maybe admin- maybe a supset of members so inclined??) to quickly and easily "pull up" all those watches (both known and unknown) which meet the requirements, as per you post, and see all these "outliers" in one feld swoop. Then some rule for dating could be discussed, based on the ones that don't fit the pattern listed, and these records would all be in one place/file etc... They are really the interesting exapmles anyway, as you and others have said.
Greg spent considerable time and effort looking at his subset of watches, and described his findings nicely- but it took much more time and effort than could be needed.
I was thinking there is so much useful "information" available which is now clunky to extract.
Wouldn't it be nice to "see" a single post (or PM to those interested) with the links to each "exception" or a summary of what the exceptions were?
Seems like this would facilitate rulemaking and allow the discussion to be based on all of the exceptions with less effort and more confidence.