Monday, January 5, 2015

Seasonality of portable typewriters

Over several decades the portable typewriter is advertised as a Christmas gift. Starting from at least the twenties right through to the end of the typewriter era, there are full page Christmas advertisements for typewriters. (A very large and illustrative selection was posted last December on Rob Messenger's oztypwriter blog.) These promote the portable typewriter as a gift for private home use.

With the impressively expanded serial number page for Remington at the typewriterdatabase, there are monthly serial numbers for some of the portable machines. E.g after September 1928, there is a consecutive serial numbering for the Remington Portable typewriter, a type that is advertised as a Christmas gift.

To what extent was the portable typewriter actually a gifting item then at that time. Did the boy indeed get it? If the portable typewriter was significantly a gifting item, then there should be some visible seasonality in the production numbers.

Taking the serial numbers for the Remington Portable from the database and plotting the size of the monthly blocks as a volume yields the graph below. This suggests that at least for 1929, the monthly increments were done by choosing every month a starting number. The monthly volumes are suspiciously nicely rounded numbers; 1000 or 1100. After summer 1930, the numbers start to suggest consecutive numbering, i.e. they probably more closely represent monthly production volumes. (The sharp drop in block size in the summer of 1930 is a bit too late to attribute it to a likely sharp drop in volume sold after October 1929.)

Looking at the overall chart, there seems to be no clear annual pattern in production volume.

Even when allowing for a longer time from manufacture to actual consumer purchase than there is today, this would surely not be longer than 12 months. The distribution and warehousing also then would have been able to route a machine to the purchaser within no more than a few months. A typewriter was too expensive an item to make to stock, would take up too much capital. 

There is perhaps a suggestion in the graph that most are produced in the first half of the year. That could then mean that time from production to sale was around 6 months and the majority of machines is sold for Christmas. An alternative explanation would be the start of the school season being a driver for purchases, with a couple of months lead-time from production to purchase.

The numbers for 1933 however are concentrated in the second half of the year, making it less likely that there really is an annual pattern. Also the volumes become relatively small after 1931 for this type. Overall the graph does not give much insight into the effect of advertising the machine as a Christmas gift. It looks a bit random. At most it could be concluded that there is no clear indication of seasonality in the serials for the #2 and #3 Portable.

With the #2 and #3 numbers trailing off, perhaps another machine would be likely to be a home gifting item. E.g. the Compact Portable of that period. Taking its monthly serial numbers from the database and plotting the size of the monthly blocks gives the graph below.

Again there is no clear annual pattern in the monthly volumes. It may be that the graph is shifted by one (first or last of month), but even then there is no consistent production peak correlating with December.

Ergo the gifting of a typewriter for Christmas may have caused a surge in sales for Christmas, but the monthly serial number blocks do not show effect of this in annual production patterns.

Not sure if the boy did get it...

(A very impressive resource is the typewriterdatabase. As already remarked by Ted upon his expanding the Corona page, the data can contain answers and hints for many more and more varied questions than the dating of a machine.)


  1. This is a good effort to start using "big data" for our understanding of typewriters.

    A difficulty in analyzing Remingtons is that in the 1930s there was an explosion of different models, as they tried to meet every possible source of demand in a challenging market.

  2. Also, something I noticed in the Corona 4 production is that manufacturing of parts (and stamping the serial number) can happen many months before the machines are actually assembled, painted and distributed. Corona seems to have tracked "last serial manufactured" (ie: frames built but not assembled). it's probable that Remington tracked distribution numbers (Sheridan was a marketing manager at Remington and would likely have had records for last serial number *distributed to dealers*). It's possible that production orders for Christmas were filled much earlier in the year for distribution.

    It's notable that the numbers for the Corona 4 show that Corona had a lot of problems at the end of 1931 as the Depression deepened and demand plummeted. They had to cease production and spent most of the next two years assembling parts made in the previous year. It seems likely that Remington had similar discord between planned production and actual sales.

    It's great to see the numbers get crunched for the "bigger picture" of what was happening in historical context. I think Christmas 1940 (ok, maybe not, as WW2 was brewing - it's tough to separate history from production) or 1950 might crunch out to give the result you expected. (:

  3. What software did you use for the plots? The look is much more sophisticated than what I've been able to produce with Microsoft's Excel. Must be an Apple tool, no?

  4. Getting all these numbers and shared worldwide is something really new. Opens up yet another interest/flavour in typewriters :-)

    Their production in the 30ies does seem a bit haphazard. Batches of varying types, rather than steady output. Looking at the fifties may indeed make more sense. By then the manufacturing would have been set up with a more efficient, linear flow.

    Even though the US demand by '40 should be still high, export would have started to suffer. The effect of the educational orders and then the actual orders for the war effort on the factory efficiency would be there by the fifties and overall volume will also have been higher.

    Am amazed that Corona stocked numbered frame parts and then assembled. The cost and the hassle of doing a fifo buffer in an otherwise batch factory. Wonder what the purpose of the serial number was from factory perspective - unique ID (warranty) and/or dating (warranty, re-sale value).

    The plots were indeed done with Numbers :-)
    But i did some manual graphics tweaking also to emphasize the years. (Tufte would disapprove...)