[And now, Part 4 of this ongoing interview with Robert W. Harwood about his book I Went Down To St. James Infirmary. Part 1 is here, and Part 2 is here, and Part 3 is here.]
Q: I want to get back to the legal wrangling, which was among the most eye-opening sections of I Went Down To St. James Infirmary, but first a quick digression.
You hipped me to this post detailing Louis Armstrong’s various takes on the song. Vis a vis the 1928 recording, the author remarks that the “record must have sold quite well because a score of ‘St. James Infirmary’ records hit the market in 1929 and 1930.” What do we — or can we — know about sales of that recording at the time? Was the Armstrong a “hit,” as they say?
A: One of the appendices in the book details the versions of the song that were released between February 25th, 1927 and December 23rd, 1930. Armstrong’s version was recorded on December 12th, 1928, but wasn’t released until February of 1929. The Primrose copyright was registered on March 4th. The next versions didn’t appear until October — by The Hokum Boys — and then three more recordings were released the next month, including an instrumental version by Kansas City Frank. The song was really heating up the airwaves by this time, and seven or eight new versions appeared in January of 1930.
This was when Denton and Haskins released the sheet music labeled “St. James Infirmary or Those Gambler’s Blues also known as St. Joe’s Infirmary.” [Below right.] Mills launched a law suit and simultaneously copyrighted — again as Joe Primrose — a new “St. James Infirmary,” as a variation on a traditional song, with new lyrics: “Tho’ she treated me mean and lowdown, Somehow I didn’t care, Now my soul is sick and weary, I hope we’ll meet again up there.” I think he did this as a cushion, in case Denton and Haskins won their right to market the sheet music. One of those January releases was by The Ten Black Berries (aka Duke Ellington and his Orchestra) with Mills himself — as “Sunny Smith” — handling the vocals.
Here I have to note that racially integrated music groups didn’t exist yet, although Mills had tried to put some racially mixed groups together in the recording studio. Part of the reason was discriminatory, but it was also due to the record companies insistence on classifying their releases; there was no category for this.
Anyway . . . Mills’ group, Mills’ Merry Makers, also released the song, with a sixteen-year-old Jack Teagarden handling the vocals. I remember, Rob, your comments on this version in Letters from New Orleans: Jack “delivers a take that works so hard to get the verb tense right that it sounds like a grammar teacher delivered it” and he thereby “misses the mysticism and the nastiness” in the song. That’s good. Both these versions — The Ten Black Berries (using the Armstrong/Redman arrangement) and Mills’ Merry Makers — were based on the second copyrighted version.
Eight more recordings came out before the end of 1930. I mean, this song was really hopping! By March of 1930 — one year after Armstrong’s version hit the streets — over 37,000 copies of the sheet music had been sold by Gotham Music Service, Inc. (the publishing arm of Mills Music). It was a valuable commodity, and continued to gain value. Within two years of its release it had sold over 200,000 records and was considered one of the hottest sheet music sellers in the country.
These were really big numbers for the 1930s. It was the start of The Depression. Victor’s total record sales had dropped by over 50% between 1929 and 1930, and in 1931 were a third of the previous years’ sales. But “SJI” was like a galloping race horse, nothing seemed to impede its momentum. Armstrong’s version remained the model, and was undoubtedly one of the biggest hits of those early Depression years.
That’s interesting. I’ve always had trouble gauging the “hit” level — I knew enough about the way the “song plugger” stuff worked, and the sort of overlap between sheet music and actual record sales as revenue streams (I think that transition was in its final phase by this time), and it always made wonder if a lot of versions was a sign of a hit, or just a sign that something was being pushed. So I’d never never seen sales figures before — there wasn’t like a Billboard Top 40 in those days, was there? Was it hard to locate that data, number of records sold, sheet music sold?
I would love to get hold of sales sheets for OKeh around this time. I’ve got no idea if they exist, though, and no idea how to locate them if they do. I got most of those numbers from a 1930s court case, where lawyers for Gotham/Mills Music were showing what a valuable commodity they had in “St. James Infirmary.”
All those variations of “SJI” that emerged at the end of 1929 and then through 1930 were definitely a sign of its popularity. Not everybody could afford Armstrong’s recording of the song. Many of the alternate versions were marketed through budget labels. The two released by Duke Ellington, for instance, were for budget labels. These would go for 25 or 50 cents a record, as opposed to around a dollar for a “premium” recording like Armstrong’s OKeh release. This is why Ellington used pseudonyms — if he released premium recordings and also budget recordings under his own name, he would be in competition with himself; sales of his more expensive records would slip.
So, as The Ten Black Berries, the Ellington orchestra released three separate takes of “SJI” (all recorded on the same day) on a number of budget labels including Romeo (sold exclusively in the S.H. Kress & Co. department stores), Oriole (sold through the McCrory store chain), and Conqueror (sold through Sears, Roebuck). These budget releases also included records made from a flexible “Durium” material (a kind of shellac-covered cardboard) and sold from newsstands throughout New York City for about 15 cents. The public didn’t know who a lot these performers were; it was the song that was the draw.
Mattie Hite did a fine version of “St. James Infirmary” — which she called “St. Joe’s Infirmary.” She recorded it in January of 1930, so this was in the very early days of the song’s recording history. “SJI” had already made a mark, though, and I think people bought her record because it was “SJI,” not because it was by Mattie Hite. Hite did not record many songs, about eight altogether, and while she was something of a local draw she did not have an established reputation. Much is made of Mattie Hite’s recording. I’ve read that she had a chance of making herself better known through this song, but was knocked out of contention by the popularity of Cab Calloway’s release. I really have my doubts about that. Almost a year (and ten other versions) separated Hite from Calloway. Hite’s version contains eleven verses, it clocks in at three minutes (the maximum that a 78 rpm record could hold) and she sounds just a little chipmunk-like, as if it had been sped up in order to fit it all on the record. Anyway, she was one of many lesser-known or pseudonymous performers who recorded the song in those early years. Although stars such as Gene Austin and King Oliver also had versions on the market. It would be really interesting to know how well these individual versions sold, but I don’t know of any surviving record for that sort of information.
But, Rob, to get back to your question: How big a hit was SJI in, say, 1929 or 1930? I don’t think we can know for sure. We can deduce, based upon the available evidence, that it did very well indeed. But they weren’t collecting music sales data then as they are today; there were no Soundscan or Billboard charts. Billboard had been around for quite a while, as an entertainment magazine, but it didn’t post its first chart until 1936 (listing the most-played songs on several radio networks) while its first chart of best-selling records didn’t appear until 1940.
In those days, as you said, its popularity would have to be gauged through sheet music sales as well as record sales. Sheet music sales of 37,000 copies in a year would have made it a valuable bit of merchandise, but not a runaway success. I’d guess that Armstrong’s release of the more socially acceptable song “When You’re Smiling” at around the same time sold more records. But I’d also guess that “When You’re Smiling” did not generate the numbers of budget recordings that “SJI” did. If we could add up the total sales of the twenty recordings released in 1929 and 1930, how would those compare with sales of Armstrong’s version alone?
Having said all that, Rob, I’ve just run across a book that lists the top pop songs from 1890 to 1954. This is an ambitious project, and many of the lists must be educated guesses. Still, Joel Whitburn’s Pop Memories 1890-1954: The History of American Popular Music places Armstrong’s SJI at #15 in 1929, King Oliver’s at #9 in 1930, and Cab Calloway’s at #3 in 1931.
Maybe someone who reads this will know the actual numbers of records sold. I’d sure like to hear from them.
[…] Irving Mills (himself disguised as “Sunny Smith).” That version mentioned a bit in this Q&A with Robert W. Harwood and in my SJI […]