Louis Armstrong’s ‘Cornet Chop Suey’ (1926). What key is it in?

This article could not have been written without the generous help of Michael Kieffer, to whom many thanks. Other acknowledgements will be found at the foot of the text.

Over the years, I had occasionally heard that some doubt existed as to the correct key for Louis Armstrong’s tune of this name. The doubt specifically concerned the original version of it, which he had recorded with the Hot Five for OKeh early in 1926. This problem had apparently been around for some years. It had been discussed in the correspondence columns of Jazz magazines; possibly articles had been written about it, and it had certainly been talked about quite a bit. I understood that well known trumpet players had gone into the problem, and that, surprisingly, there was still no general agreement.

A few years ago, I became interested in selecting the correct pitch for early Jazz and dance band records, and found that by applying a few simple tests, it was – usually – possible to be fairly sure of the correct speed at which to play a 78 rpm record, so that it would come out at the correct pitch.

However, these tests were only valid for Jazz and dance records made in the U.S.A. and Britain in the 1920s through to about the mid-1930s; and even then, only when the band included a piano.  It primarily rested with the piano, of course, and the assumption that this would be tuned to a standard pitch. I asked the late John R.T. Davies, the doyen of 78 rpm record restorers, whether this assumption was acceptable. He agreed strongly, pointing out that the major record companies (Victor, HMV, Columbia, Brunswick, Vocalion, Odeon, OKeh &c.) were large concerns, recording the most prominent international artistes, and the use of first-class pianos was to be expected, and therefore, for pitching purposes, that assumption was valid, tenable; indeed, unavoidable.

Of course, there are instances of ‘below-par’ pianos to be found on some Jazz and dance records of this period. However, these are probably pianos that are simply rather out of tune with themselves, and sound ‘ploingy’ as a result. This is quite a different thing from the piano being tuned to the wrong pitch altogether. (See appendix 1.)

So in general our assumption that the pianos are tuned to standard pitch is valid as a starting point. In any case, if for example, a piano had been allowed to become very flat in pitch, it would be difficult for wind instruments – the clarinet in particular – to ‘get down’ to the pitch of the piano without becoming out of tune with itself. And if a piano had somehow been tuned very sharp, a clarinet would simply not be able to get up to that pitch at all. Overall, the statement: ‘Pianos in recording locations, whether permanent or temporary, were, in general, tuned to standard pitch’ is a reasonable one, and likely to be true far more often than not.

And what actually is this standard pitch? As far as the U.S.A. goes, the note A (the one above middle C on the piano) should be 440 Hz, usually written as A=440. And the standard pitch used in Britain for orchestral and dance music at that time (circa 1900 – 1950) was A=439, a trivial difference, so that the same tests can be used pretty safely for both countries. (See appendix 2.)

As for other countries, and other styles of music, and indeed those artists and ensembles in the U.S.A. and Britain not using a piano, the application of ‘The General Rule Of The Piano’ must – in the first instance – be assumed to be inapplicable and, consequently, conclusions from it non-viable. I am not qualified to comment further on these musics; but certainly commend those who may be interested in them to pursue their own researches on these fascinating topics. Perhaps they will be able to derive some simple tests to help ensure correct pitching of old 78 recordings of e.g. a Javanese gamelin orchestra, or a Cantonese instrumental ensemble? After all, the correct pitching of any and every ‘78 rpm’ record is an essential part of properly preserving, for posterity, the information contained on it.

About three years ago, I heard of the existence of a CD set of early Louis Armstrong classics that included the 1926 Hot Five ‘Cornet Chop Suey’ twice. Once in the key of E flat; and also in the key of F. This was because, in the opinion of the compilers of the set, there was still no general consensus on which key it was in. To include it, therefore, in both keys was certainly very commendable. But I was puzzled that a record could be attributed to two keys so much as a tone apart. Not merely a semitone, but a whole tone: really a very large interval! In theory at least, it should have been fairly easy to decide which was the true one. The trumpet players who disagreed on the key of the piece may have (I don’t know) played the tune over on their trumpets (or cornets) in both keys.  And then used, as a basis for their conclusion, the fingering of their horns indicating one key rather than the other because one key  ‘fell more naturally under the fingers’ than the other.  At least, I assume that this is what they did. If my assumption is correct, then I have to say that that approach might at times be deceptive. As a clarinet player, time and time again, I have tried to find out exactly what Johnny Dodds or Don Murray played on their clarinets back in the 1920s, and the more I learn, the more I distrust what seems logical on the surface. Also, as the decades pass, it becomes ever more difficult to even attempt to analyse the ‘mindset’ of a 1920s virtuoso player. Certainly, Dodds and Murray were both virtusosi of the clarinet. They could play anything they mentally conceived… and usually did so. Perhaps intuitively, they ‘eliminated the instrument from the equation’: the music that appeared in their consciousness was the music that straightway sounded in the club or the ballroom in which they were playing. There was no intervention of any ‘problem of execution’ on their instrument. If – as I suspect – they (along with most other top musicians) did this, they were rather in advance of their time. They did not need to read treatises on the psychology of musicianship, the bulk of which have proliferated in the last 50 years. They just did it anyway.

If Dodds & Murray could do that, how much more could Louis Armstrong do it? Louis, from his first startling appearances on disc in 1923, was manifestly a very special case. On this basis, Louis’s cornet fingering patterns, I thought, might be rather unsusceptible to logical analysis.  I’d found exactly the same in trying to play Dodds’s clarinet solo on ‘Potato Head Blues’ by the Hot Seven on a clarinet in C, in case he was playing one of those, instead of the normal B flat clarinet. Both fingerings, I found, were pretty equally plausible.

It was decided to investigate the problem of ‘Cornet Chop Suey’ from a different angle. Indeed, a more technical or ‘scientific’ angle, in the hope that this type of investigation might be more secure in its findings.

Fortunately, ‘Cornet Chop Suey’ does not exist in isolation. It is one of six titles that were recorded at a single session in Chicago on 26th February 1926. A very important session indeed, as no less than 4 of the sides became classic Jazz standards. (‘Oriental Strut’ and ‘You’re Next’ tend to be the rather neglected ones.) Here’s what they recorded that day:

9533-A      Georgia Grind                       OKeh 8318
9534-A      Heebie Jeebies                      OKeh 8300
9535-A      Cornet Chop Suey                OKeh 8320
9536-A      Oriental Strut                        OKeh 8299
9537-A      You’re Next                          OKeh 8299
9538-A      Muskrat Ramble                   OKeh 8300

The obvious thing to do was look at the keys of the other 5 sides on the session, and check for anomalies. There was clearly something curious going on, as there had never been, to our knowledge, any question or dispute about the pitch of these other sides. Why should ‘Cornet Chop Suey’ alone be the cause of a problem?

In order to check the keys of the performances it was essential to have access to original master pressings. These, of course, are exact replicas of the wax blank on which the record was cut. They bring us, in one single step, right into that studio in Chicago, and to the cutting lathe on which they were made. The use of any other version, e.g. a dubbing (transcription) onto another 78, tape, LP or even a CD brings with it the risk that the pitch has changed slightly and will mislead us. Happily, these classic sides were widely re-issued on 78s in the late 1930s and 1940s. These sold well and are not scarce. Best of all, four of them were pressed from the original masters. (‘Heebie Jeebies’, ‘Oriental Strut’, ‘You’re Next’ and ‘Muskrat Ramble’). This gave us 4 opportunities to check the speed of the cutting lathe. What speed did it run at? Was the speed consistent throughout the side? Was the speed consistent throughout the session?

Unfortunately, master pressings of ‘Georgia Grind’ and the all-important ‘Cornet Chop Suey’ itself were at first not available to us.

The original work on the problem was done a couple of years ago. Much valuable help and advice was given by members of the 78-l Internet discussion group. It allowed the provisional conclusions to be made, as given below. Above all, Michael Kieffer generously supplied copies of the two missing sides, transferred from the master pressings on the original OKeh discs. Now we could investigate all 6 sides at the same time, much increasing the reliability of the findings.

Naturally, all the 6 sides needed to have been transferred at exactly the same speed before we could begin the investigation.

The notional standard speed ‘78’ means different things in different countries. For many years, the only convenient and reliable way of checking the speed of a turntable was by a stroboscope. Typically, these take the form of a disc with a certain number of radial bars, like the spokes of a wheel. When placed on a rotating turntable and viewed under a light derived from AC mains, the bars are illuminated more strongly on the peak voltages of the alternating cycles, and so may appear to stand still, or progress ‘forwards’ or ‘backwards’. To design a stroboscope, we must take into account the frequency of the AC supply. In Europe it is 50 Hz. Now it is possible to have a number of bars that will appear to stand still at exactly 80 rpm, such as the example seen on the left. But it is not possible to have one that will appear stationary at exactly 78 rpm. You can only have, say, 64 bars on your disc, or 65 bars… you cannot have 64-and-a-half bars! The whole number of bars under a 50 Hz supply that will ‘stand still’ do so at 77.92 rpm. However, this error is minuscule and can be ignored. In the U.S.A. the mains frequency is 60 Hz. Again, this allows an exact 80 rpm stroboscope to be made, but not one for 78; in their case, the ‘standing still’ speed is 78.26 rpm. This error, for all ordinary purposes, is also negligible. Still, in this case, we decided that even this error should be eliminated.

Michael Keiffer’s 2 discs had, of course, been transferred by him at the ‘American speed’ of 78.26 rpm. In order to exactly synchronise our 4 sides to his, we checked our (supposedly ‘British’) record deck with its speed quartz-locked to 78 rpm. Observing our stroboscope, we found, to our great interest, that the bars progressed forward, though very slowly. Timing their advance, we found that after one minute, they had progressed about 90 degrees. There was only one conclusion: ‘quartz-lock 78 rpm’ on our deck observed the U.S. standard speed, not the British one. This at least meant that we could transfer our 4 sides without having to take any further precautions; and, as we remark similarly below, what’s a Quarter of a Revolution between friends? 8^)

This done, we matched up the general sound of all 6 sides to facilitate comparison, &c.

As OKeh records at that time were supposed to have a standard speed of 80 rpm, we would expect our sides at 78.26 to be rather flat in pitch, as they are running 2.2% slower.

We played the first side, ‘Georgia Grind’, and checked the key, using a keyboard calibrated to A=440. The performance appeared to be in the key of D. This was immediately highly suspect, as Jazz and dance bands almost never used that key. The choice of a neighbouring key was far more likely. The adjacent keys are of course D flat – occasionally used by bands – and E flat, this being a frequently used key. As we knew our sample was already on the flat side, we immediately opted for shifting the performance up into E flat, and taking it from there.

It was necessary to raise the pitch by 5% to bring the music into E flat. If this key was in fact valid, and if the cutting lathe was consistent during the session, the same correction should bring the other sides into line.

Next came ‘Heebie Jeebies’. Now there is a universal consensus between the bands and musicians who play it that this in the key of A flat. But the sample played more or less in G. This is a semitone lower; it also happens to be a key that was used quite often in this period, indeed more often then than now. But we followed our preliminary findings, and raised the file 5% as before, putting the track into A flat – the universally accepted key for it.

Now for the main track: ‘Cornet Chop Suey’ itself. This played more or less in the key of E. This is obviously between E flat and F, and while those 2 keys are very frequently used, E major itself was virtually never used in normal Jazz and dance music. Had anyone taken a copy of OKeh 8320 and played it at 78, the problem would have presented itself immediately: whether to take it up to F, or down to E flat? This is almost certainly how the problem first began. (See Supplementary Comment below).

As is now emerging, these sides were recorded quite fast; to raise them to plausible keys we have to increase the pitch by 5%. As the relationship between groove speed and pitch is linear (if we double the speed, we increase the pitch by one octave), we can easily establish the speed of the cutting turntable that day in 1926. It was 5% above 78.26 rpm, which is 82.17 rpm. OKeh were actually cutting that day 2.7% above their own recommended standard of 80 rpm.

However, we still have three sides to consider.

‘Oriental Strut’ begins at 78.26 rpm in the extremely unusual key of D flat minor, resolving to the ‘forbidden key’ of E major. In fact, if this were the true key, the opening verse should actually be designated C sharp minor. This is utterly preposterous for a Jazz outfit; but raising the pitch by 5% brings all into the very sensible keys of D minor and F major. Moreover, the author has played this very appealing and much-neglected tune on a number of occasions, when those were the keys employed, and which have universally been accepted for this piece for many years.

Again, the very little-known ‘You’re Next’ begins with an impressive ‘classical’ piano introduction, and Louis plays the verse, which – at 78.26 rpm – is in the key of E minor. Such a thing is highly implausible, and although the chorus is in the key of G, everything is much better when the performance is re-pitched up a semitone. Then, the chorus is in the common key of A flat, preceded by its familiar relative minor, F minor. These are much more likely keys.

Finally, the classic ‘Muskrat Ramble’ plays in G at 78.26. That may well be a viable key for it, but by now I think we have shown that this entire session needs to go up 5% from 78.26 (or 2.7% from 80, whichever way you prefer to look at it), and so ‘Muskrat’ turns out to be not in G, but in A flat, always a key to be favoured by trombonists, we understand! (Curiously, when Kid Ory, the composer of ‘Muskrat Ramble’, began playing again after his retirement during the 1930s, he seems to have moved ‘Muskrat’ from A flat to B flat, which is the key in which it is now universally played. Also, he then played it in the form A-B-B-A-A….., while the Hot Five version is A-B-B…..A. But that is a mere detail.)

Enough! We are guilty, perhaps, of using a sledgehammer to crack a walnut…

Little remains to be said, except that on that day in February 1926, there was another feature of the OKeh cutting lathe. All the above discussion has been based on audio samples from near the beginning of the sides under examination. If the various keys are again checked near the end of the sides, they are actually slightly more flat than they were to begin with. For example, re-examining the end sections of the first and last sides of the session, we find that a pitch increase of nearly six per cent, rather than five, is necessary to restore them to the correct key. This means that, during the course of cutting the side, the lathe actually increased its speed. This is easily accounted for. The lathe should have kept exactly the same speed throughout, of course. But, if there were any deficiency in the torque applied to drive the turntable, we would expect it to be slower near the beginning of the side, when the drag of the stylus cutting the groove in the wax would be appreciably greater due to the high groove velocity. Near the end of the disc, the groove velocity is much less, and so is the drag from the cutting stylus… thus, the turntable is able to speed up somewhat. This speed change during the cutting of the discs is consistent for all six sides.      

It’s time to sum up.

Our only assumption in this matter is that the piano in the studio in Chicago was tuned to A=440, for the reasons stated, at some length, above. Therefore, on this session of 26th February 1926, we suggest that:

(a) The cutting lathe was running rather fast that day.
(b) At the beginning of each side, it ran at ~82.2 rpm.
(c) By the end of each side, it was running at nearly 83 rpm.
(d) The cutting lathe behaved consistently throughout the session.
(e) The resulting masters, when played at 80 rpm would be rather flat in pitch.
(f) If played at 78 rpm, they would be half a tone flat, a grievous discrepancy.
(g) In particular, ‘Cornet Chop Suey’ was performed in the key of F major.

It only remains to give links to two versions of ‘Cornet Chop Suey’ made from the original OKeh kindly sent by Michael Kieffer, so that you may hear & compare them for yourselves. The first is in the true key of F; the second in the false key of E flat. Of course, it may be that you prefer the slower E flat version. This is perfectly acceptable. One is entitled to hear – and above all enjoy – music exactly as one wishes. It would be tedious indeed, if, before playing a record, we had to have a Great Inquest into its exact pitch. Besides, if recordings are played slowly, it is much easier to analyse them ‘on the hoof’ and listen to what the musicians are doing; this is a valid technique for study and enjoyment. So everything has its place!

‘Cornet Chop Suey’ in F (the true key)


‘Cornet Chop Suey’ in E flat (the false key)


Supplementary Comments. 

To be fair, much of the above already seemed obvious. For example, Louis later recorded ‘Cornet Chop Suey’ unmistakably in the key of F. But, this was not sufficient to indicate that his original rendition must also have been in that key. And that is fair enough: we have already observed that Kid Ory later moved his ‘Muskrat Ramble’ from A flat to B flat; so why might not Louis have moved his composition from E flat to F?


Actually, the undoubted cause of all this confusion, was what we might call ‘The Rogue Dub’ of ‘Cornet Chop Suey’, shown above. This was made in 1937, by the American Record Corporation, (ARC), as a custom pressing for the Hot Record Society. The original 78 – OKeh 8320 – is evidently quite rare, hence its selection for reissue at such an early date. Now although ARC usually made very creditable custom dubbings of material in their archives, in this case for once they did not. The Canadian expert David Lennick kindly pointed out that this dubbing varies in speed, first in one, and then in the other direction, as it is played. This is exceptionally misleading, and must have started the whole puzzle. If anyone attempted to determine the pitch of ‘Cornet Chop Suey’ from this HRS dub, it would have been fraught with problems. As a matter of fact, correct pitching of 78 transfers was a much-neglected topic until quite recently. In the early days of the Jazz Revival, its proponents were very long on enthusiasm, but understandably rather short on experience and technical know-how. Notice that the banjoist on the HRS label is given as ‘Buddy Sincere’. This is probably a combination of the banjoists Buddy Christian (who it isn’t) and Johnny St. Cyr (who it is). Still, the record did come out, and people heard it and loved it, which was the main thing. When Columbia re-emerged under the aegis of CBS in 1939, they inherited this recently made ARC dub. It bears the master number P-21296. When CBS commendably began to reissue the Louis Classics in their albums, they used some OKeh masters, and some dubbings. They did not make a new dubbing of ‘Cornet Chop Suey’; why should they? They already had P-21296, which had only been made just a little while before. So, naturally, they used it on U.S. Columbia 36154. Later, other affiliates of CBS used it too. No less than five issues of it were made in Europe on different Columbia labels. I know that British Columbia DB-2624 used the misleading ARC dub. I’m pretty sure that the French (BF-474), Italian (CQ-2125), German (DW-5069) and Swiss (DZ-540) all used it too; but confirmation of this would be welcome; credit will of course be given here. Note: In July 2009, Dr. Martin A Lobeck of Germany kindly confirmed that German Columbia DW-5069 is indeed pressed from the master P-21269.

The upshot of all this, is that bands & musicians in the U.S.A. and much of Europe, would for many years only have had the ‘rogue dub’ on 78 to establish the true key of this piece. And subsequent microgroove vinyl issues of it did not help matters. It would seem that very many of these EP, LP and even CD reissues were still made from the ‘rogue’ 78 dub P-21296 rather than from a decent copy of Okeh 8320.

No wonder people got confused!

Appendix 1.
Two examples of questionable-sounding pianos that come to mind: (1) the piano in the ‘Ambient Studio’ used by OKeh in from early 1927, probably at 11, Union Square, New York. Bix Beiderbecke recorded his classic ‘Bixology’ (‘In A Mist’) on this piano. It was not one of the world’s best pianos, and presumably is the same one he also used on ‘For No Reason At All In C’ and ‘Wringin’ & Twistin’’; on these last two, Frank Trumbauer apparently has no problem in getting in tune with the piano (or indeed Bix when he makes his contributions on cornet) so the piano is probably in standard pitch. True, cornets and saxophones, being ‘outdoor instruments’ have a much wider tuning range than e.g. the clarinet. (2) Again with OKeh, but this time in Chicago. Earl Hines made 4 issued piano solo sides in December 1928 (‘Caution Blues’, ‘A Monday Date’, ‘I Ain’t Got Nobody’, ‘Fifty-Seven Varieties’) on a piano that would have benefited greatly from some attention. There are doubtless many other examples. But Jazz, dance music, popular vocal, instrumental and ethnic records were always the way in which record companies of those days raised the money for their prestigious classical and operatic recording ventures. So we can scarcely condemn them for occasionally providing a ‘below-par’ piano for what to them was an inconsequential ‘Jazz Band’, in view of the astonishingly enormous legacy of Jazz & dance recordings they bequeathed to us!

Appendix 2.
This seemingly evolved from the use in America in the nineteenth century, of the ‘Diapaison Normale’. This was the French standard pitch of A=435 Hz. All was fine for quite a while. However, this pitch had been established for an ambient temperature of 59° Fahrenheit (15° Centigrade). French woodwind instruments, e.g. oboe, flute, clarinet, bassoon, cor anglais, were designed to conform to A=435 at 59° F. These instruments are, mostly, not capable of much variation in tuning. All – with the partial exception of the bassoon – are primarily ‘indoor’ instruments where wide ranges of pitch are not – or should not be – required. The use of French woodwinds, or American woodwinds made to the French pitch, ran into trouble in the later nineteenth century in the U.S.A. Rich patrons of the arts built and endowed new concert halls and recital rooms, and these, of course, were ‘steam heated’ or centrally heated, in accordance with modern American practice, for the much colder winters experienced in the north of the U.S.A. These halls were typically heated to about 72° F (~22° C), and this caused chronic tuning problems for the old French standard woodwinds. They became much too sharp in pitch at this higher temperature. Eventually, from around the turn of the nineteenth century, a new higher standard pitch of A=440Hz was progressively adopted in the U.S.A. This gardually became, by the mid-twentieth century, the Standard World Pitch for ‘Western Music’. Incidentally, Britain had, by a separate (but happily convergent) process, adopted a standard pitch of A=439. This was close enough to A=440 to not cause any serious difficulties. For example, although American clarinets were not noticeably imported into Britain in the 1920s, American saxophones – the icon of the Jazz Age! –  did come over, and in vast numbers! There was – and is – no problem whatever in playing A=440 saxophones at A=439. I played a 1921 Buescher tenor sax, and it was no problem to get in tune. After all, what is one cycle per second between friends? 8^) 

Acknowledgements: My thanks to: Michael Kieffer, David Lennick, Mike Thomas and the late Jean-Pierre Lion. (If I have left anybody out, please tell me.)


© Norman Field, October 2005.

Re-formatted & slightly revised in November 2017, this page has been allowed to remain here, even though it is excessively verbose. It was one of my earlier efforts; please forgive me if I regard it with some affection. Later, Mark Berresford did further research, and discovered that from Chicago, the OKeh portable gear was taken directly to St. Louis, where more recordings were made. By then, the lathe was going even faster, so a still larger correction must be applied to the sides recorded there.