... I reject entirely a proposition I have seen that 88 and 88K production overlapped. The estimated year boundaries changed very little while the sample increased from fewer than 20 to 80+, so we can pretty safely say that sampling error is small now..
It was a good statistical story, until a new bit of data intruded (assuming this one-off is not itself a worker error). I have discovered an 88K with serial 1848292, whereas the latest 88 is 1916010. That is, in 1953 there was an overlap of about 70,000 pens when both 88s and 88Ks were produced. I have three 88 samples and one 88K sample within that range. Other than obliterating my "entirely reject", this makes very little difference to dating of either pen model. Far more important is the following hitherto neglected item.
Second anomaly is in the 88K series. I have serials from straight after the last 88 (about 1.9M) to just under 2.2M then nothing at all until just over 2.5M when a fairly uniform distribution resumes. While this could be chance, was there a model tweak during the run of the 88K?
After answering a PM enquiry about dating an 88K, I decided to spend some more time trawling ebay for evidence of serials for this model to see whether I could resolve this anomaly of which I had been reminded when looking at the spreadsheet.
With double the number of 88K pen serials now available, the anomaly remains. I conclude that the face of it is the fact of it; on or before number 2.2M they skipped, re-starting serialisation at 2.5M, within the 88K series. Why? I do not know. I know that of three Aquilas I have on record, all are serialised above 2.5M but so are several plain 88Ks, while an 888 is back in the low two millions. Did they change factory? Machinery? Design detail? We know they were making some changes to materials along the way, first of the barrel then of the piston knob and section. There is no evidence of a change of dimensions until the 88P.
One effect of incorporating this into the data is that apparent production of the 88K model drops with the missing serials. I now estimate only about 420,000 of that model were made. Consequently, year estimates shift for the 88K. I now have the following boundaries for serials (rounded numbers):
1953 - up to 1.9M
1954 - under 2.0M
1955 - 2.1M
1956 - 2.17M (up to 2.2M with the skip on the year boundary to 1957?)
1957 - over 2.5M, possibly including about 30,000 below 2.2M
I estimate around 100,000 pens produced in 1957, with the 88P appearing for 1958 with a starting serial of 3M.
Sometimes I write with pens, sometimes I fix them, and sometimes I manage to extract fun from looking at serial numbers.
"...all observation must be for or against some view if it is to be of any service." Charles Darwin