The 45 rev/min “EP” was probably commercially driven. Perhaps it was an exercise in how to get a bit more out of the 7-inch 45 rev/min disc than was originally envisaged. Recording at lower average levels, with some compression might have been used in order to get closer groove spacing. I don’t know if RCA issued a paper or technical article on this development, which was c.1952.
The original 45 rev/min work was well-documented in an article in RCA Review for 1949 June. This may be found the excellent American Radio History site,
https://www.americanradiohistory.com..._Issue_Key.htm. I have extracted the article, which runs to 18 pages, and is entitled “A Record Changer and Record of Complementary Design”.
Attachment 172944
The derivation of the rotational speed was quite interesting. For the changer to function relatively quickly, that is within one revolution of the record, the maximum rotational speed was 1.2 rev/s, or 50 rev/min. Allowing a working margin suggested a speed a little below that, namely 45 rev/min. It would appear that the earlier notion that disc rotation speed should be the quotient of integral division of 1800 rev/min (the speed of a 4-pole synchronous motor operating on a 60 Hz supply) was still in vogue. 78.26 rev/min was 1800/23, 33⅓ rev/min was 1800/54, and now 45 rev/min was 1800/40.
RCA then did the calculations which showed that 45 rev/min was indeed a better choice than either 33⅓ or 78.26 rev/min, the latter also introducing complications from the record changer viewpoint. Minimum volume of material for the desired playing time was one of the objectives. Another was 5⅓ minutes playing time with a terminal linear velocity of 11.5 in/s. In fact the disc dimensions allowed 7¼ minutes playing time at a terminal linear velocity of 10.0 in/s, which was acknowledged as producing lower quality. I’d guess that the lower terminal linear velocity was used for EPs.
A comparable article on the LP was probably provided an article in the IRE Journal for 1949 August, now available (for purchase) as an IEEE paper, at:
https://ieeexplore.ieee.org/document/1698116, “The Columbia Long-Playing Microgroove Recording System”, by P. Goldmark et al.
I haven’t seen that paper, but I imagine that it includes the rationale for the 33⅓ rev/min speed choice, beyond the fact that this was an established transcription record speed. I’d guess that it was taken as an “in the ballpark” starting point, with subsequent calculations confirming that no departure from it was materially beneficial.
Cheers,