My research has only been able to take me as deep as The Chemistry of Penicillin book, a tremendous collection of the vast literature and information on Penicillin research.
Quoting from the p. 54 of the book (Chapter IV - STATUS OF THE RESEARCH ON THE STRUCTURE OF BENZYLPENICILLIN IN DECEMBER 1943):
When the results of analyses of crystalline sodium benzylpenicillin were found to be in agreement with C16H17N2O4SNa (Chapter V), it became evident that this penicillin gave an isomeric "penillic acid." Therefore, the penicillin precursor of the "penillic acid " obtained by Duffin and Smith should have the formula C14H20N2O4S. The existence of at least two penicillins was therefore established, and the system of nomenclature using the letters G, F, etc., was introduced (Merck, M.1, 1). The name
"penicillin G" is now replaced by benzylpenicillin, "penicillin F" by Δ2-pentenylpenicillin, etc.
That Merck M.1 reference is a bit elusive, I'm afraid. The book's Appendix states that (p. 1056) the M.1 report of Merck & Co., Inc. has a "Date of Report" Nov. 1943, as well as that it was received by the O.S.R.D (N.Y.) on 28 Dec. 1943 and by the Medical Research Council on 14 Feb. 1944.
The closest I have been able to reach, regarding the M1 report is this UPenn library link, which has a number of very interesting reports.
I have started with the Penicillin: brochure with annotated bibliography which has a full view here. An interesting reference in there points us to (p. 8):
Shortly thereafter Roberts and his co-workers486 also isolated an antibacterial substance distinct from penicillin, to which they gave the name penicillin B.
For convenience we shall refer to our product as penicillin B to differentiate it from the product obtained by Abraham et al. (1,2) which we shall call penicillin A. Although penicillin B may be similar to “penatin” reported by Kocholaty (3), the absence of a discussion of its chemical properties prevents a comparison of the two products.
I am still working on it, hopefully, relatively soon, I will check back with more information.
Chasing after some references found in the Manual of the Penicillia, I managed to extract some more information relevant to the point, in The Chemical Study of Penicillin: A Brief History:
At about the same time it became clear that the penicillin which had been obtained in crystalline form in the United States was not identical with the penicillin with which the British investigators had been aorking. Among other differences between the two was the far greater reluctance of the latter to crystallize. The chemical distinction between them was clearly brought to light during the middle of 1943 by observations in several quarters. One was the demonstration by Stodola, Wach.te1, and Coghill, of the Northern Regional Research Laboratory, that the two varieties of penicillin give different, though analogous, crystalline derivatives when the free acids of the respective penicillins are treated with benzylamine. The second observation was the demonstration in the Merck laboratories, by analytical data on crystalline penillic acids, that there were at least two penicillins.1
1Systems of nomenclature using letters (in the United States) and Roman numerals (in Britain) were initiated late in 1943. However, in the interests of clarity and uniformity, workers in the field agreed early in 1946 on the nomenclature now in use, which involves a designatory prefix such as "benzyl."
This indicates that it was around the late 1943 that the names were picked. Furthermore, Coghill and Koch, (1945) recount, in Penicillin-a wartime accomplishment:
The chemistry of penicillin, as of this writing, is still classified
as secret, with the expectation that the information will be released
within the near future. One fact, however, which for a variety of reasons has now emerged, is that, as in the case of the B vitamins,
there is a multiplicity of penicillins. These compounds, all characterized by minor variations on the same fundamental structure, are produced in varying proportions on different media and by different
strains of the mold. The four principal penicillins are known in
this country as F, G, X, and K, arid in the United Kingdom as I, II,
III, and IV, respectively. When existing secrecy regulations have gone the way of other necessary wartime restrictions, these designations can be replaced by accepted chemical nomenclature.
It appears that the main purpose of letter-naming was, at first, the simple fact that, due to wartime, apparently, the exact chemical composition of the components was classified. It is, I would say, very typical of military tradition, to assign letters or cryptic acronyms to pretty much everything that is even remotely important.
I will keep the research going a little bit more, there is a vast body of literature on the subject, and it seems that every new source reveals its own multiple new sources in an endless chain reaction. I hope to have a new update relatively soon.