Nov 17, 2016

"Autumn In New York" - Comparing Charts

Vernon Duke's "Autumn In New York," a beautiful ballad that is an essential part of jazz repertoire, was first introduced as the closing number in a 1934 Broadway show, "Thumbs Up." For a nice writeup on the tune and the composer, check this article on

Charts for "Autumn In New York" can differ considerably regarding the chord changes. A couple of weeks ago I finally decided to look into what was what, and ordered a copy of the original 1934 sheet music. The sheet music really helped clarify what Vernon Duke had in mind.

Sheet music for tunes in this era consisted of the melody, printed above a complete piano arrangement. Chord symbols were generally an afterthought, added above the piano staves in case someone wanted to play along on guitar (or ukelele or banjo). The chord symbols would produce a rough approximation of the more carefully-composed piano part.

Modern charts consist of just melody and chord symbols. A lot of subtleties in the piano part may be lost, while the chord pattern may be adjusted to be more improvisation-friendly. I looked at nine relatively modern charts, comparing them to the sheet music and to each other. In this post I'll make some observations about these nine charts. I'm only looking at these printed versions - analyzing the ways that various performing artists and arrangers have harmonized the tune would be another question entirely.

The original arrangement includes a lead-in "verse" that shows up in only one of these modern charts; it's left out in most classic recordings of the tune. That's a shame, as it's quite well-written. In Vernon Duke's piano arrangement, the verse begins like this (click to enlarge):

Note the parallel motion of five maj7 chords in the first and second measures, with the first three moving by whole step - an impressionistic, Debussy-like device (Duke had studied with Gliere, and besides writing popular music, also had a career as a modern classical composer). This opening reminds me very much of the introduction to Tadd Dameron's "On a Misty Night" (1956), which uses the same parallel-harmony device. I don't think this is a coincidence.

In the excerpt above, note also the bass line in bar 5, reflecting the first notes of the familiar melody of the "chorus" (main body of the tune).

Perhaps the verse is often omitted because it's tough to sing. Here's a recording by Mel Torme that includes the verse:

The song was written with two sets of lyrics, involving a repeat of the "chorus" section. This makes for a song that was perhaps too long to be performed in its entirety on pre-LP records; I couldn't find any early recordings with both the verse and all the lyrics.

About the charts: We are considering only the familiar chorus section of the tune. In consideration of copyright, I am not showing the melody - you can find that in any fake book, and it will be almost exactly the same as the sheet music.

The charts I looked at are from these sources:
There are at least ten spots in the tune where you will find disagreements between these charts:

measures 2 (and 18), 3-4 (and 19-20), 7-8, 12, 14, 16, 20, 23-24, 26, 28.

Below are the chords as shown in four of the charts - the original sheet music (chord symbols only), Old Real Book, Hal Leonard Real Book "6th Edition," and Dick Hyman (only half of the Hyman version is shown, because IMO its copyright value is largely as a harmonic arrangement). My comments follow.

Here are the problem spots (measure numbers are for the chorus only):

Measures 2 and 18: Some charts (including the chord symbols in the original sheet music, as well as HL 6th ed., above) show Gm7, but the piano notes actually show Bb6. The Bb bass note makes for a nice ascending bass line that is important to the song. Here is a clip from the sheet music:

mm 3-4 and 19-20: As you can see in the clip above, the original harmony is static - just Fmaj7 (or F6, or F6/9 if you like). The Dm7 symbol is incorrect. The piano part fills this space with a moving line and an embellishment. In modern charts, this space is sometimes filled in with an ascending chord sequence, mirroring the ascending movement in mm 1-2 (see Old RB and HL 6th ed., above). I like Hyman's solution, with a voice-leading line (not the same as in the sheet music) specified, and Am7 D7b9 in bar 4. However, specifying voice-leading lines like this might be more detail than is necessary for a jazz chart.

mm 7-8: Originally, two II V sequences, as in the first chart above, with a voice-leading line and melodic decoration. In pretty much all modern charts, this is changed to one "long" II V, one bar per chord.

m12: In the original symbols, Cm is not really correct, I think. To me, the first two beats of m12 comes across as a continuation of the Ab chord that precedes it. The Ddim symbol is not quite right; it's actually Dm7b5 in the piano part. Old RB and Hyman get this right, but many other charts replace the Dm7b5 with G7b9. To me this sounds weak; it's an apparent effort to make everything conform to clich├ęd dominant-tonic harmony. Dm7b5 sounds better; it's what Vernon Duke wrote. Dm7b5 is II of the Cm that follows. What's wrong with that?

In mm 11-12, the "6th ed." chart takes some real liberties with the harmony, crowding this space with a sequence that doesn't work for me, and that I've seen in no other chart.

m14: For the first two beats, Duke wrote a problematic D7 to harmonize a G in the melody (see clip above). The F# in the chord clashes. Maybe he wanted a little moment of dissonance to illustrate his mixed feelings in the lyrics, "making me feel I'm home." Or maybe it was just an oversight. Many charts eliminate this clash by changing the D7 to Dm7; I agree. Hyman suggests some voice leading in showing G13 to G7+ in the last 2 beats of this measure (the original is a bit different, showing G7+ to G7 in the piano part).

m16: The last chord in the first half of the tune sets up the second half (see clip above). Duke uses C7+ on beat 4, which moves into Gm7 (actually Gm9, melody on the 9) by way of chromatic motion of the top two voices. Hyman changes beat 4 to F#m7, providing a completely chromatic approach. I like it, though it's a departure from the original score. Hyman's charts generally do tend to be more like "arrangements," and not so much "vanilla" charts. Some other charts (HL 6th ed., HL Real Jazz Standards, Aebersold vol. 40 and 93) set up the return with a conventional Am7 D7 (or D7b9, D7b5, or D7#9). To me this sounds a lot less interesting.

m20: In the original score, mm 19-20 are exactly like mm 3-4. However, some modern charts (Old RB,  HL 6th ed., iReal) fill the space as in the two middle charts above - an OK solution, but not what Duke wrote. Aebersold vol. 93 uses Dm7 on beat 3, Dbm7 on beat 4, which sounds odd to me.

mm 23-24: This is an important spot. The original piano part, and many classic recordings, have this as  | Bbm    | Bbm#5  Bbm6 | (the chord symbols in the sheet music are misleading). This line harmonizes the word "pain" in the lyrics, "Autumn in New York is often mingled with pain." Hyman keeps this progression. Aebersold vol. 93 has a sightly different line in the same spirit: | Bbm7  Bbm7#5 | Bbm6 C7#5 |. Old RB, HL 6th ed., and iReal have | Bbm6 Abm7 | Gb7   |, which doesn't work for me.

m26: The original score has this measure as | Fm Ab7+ |. HL Real Jazz Standards and iReal turn the last 2 beats into a II V, Ebm7 Ab7. This seems like another effort to turn every change into stock jazz chord movements. I don't think the chart needs this; after all, a soloist can always play a II V over a simple V, anyway. HL 6th ed. and both Aebersold charts insert a chromatic passing chord, Em7, resulting in this bar looking like | Fm Em7 Ebm7 Ab7 |. To me this is unnecessary clutter.

m28: The sheet music chord symbols show Fm7b5 on the last beat. However, it follows 3 beats of Dbmaj9, so in effect the Fm7b5 is really Db9. Note that the top notes of this chord, Ab C Eb, spell Abm, a chromatic upper neighbor to the Gm7 that follows. Great writing! If a chart expresses the last beat of m28 as either Db7 or Fm7b5, it's fine. HL 6th ed. and HL Ultimate Fake Book show Abm7 and Abm6 here respectively - not quite as effective, but following the same basic idea. However, iReal and Old RB show Ab7 (subV of Gm7), another example of changing a more interesting sound into "correct" predictability. Aebersold Vol. 93 shows D7+4, essentially the same thing as Ab7. The four-chord sequence shown by HL 6th ed. in mm 27-28 seems stilted to me.

After all that, what is the best chart? That depends on what you are looking for. For a chart that is more of an arrangement, I think Dick Hyman's is excellent. If you are looking for a basic, serviceable "vanilla" arrangement that is mostly correct, then perhaps the chart from the Hal Leonard Real Jazz Standards Fake Book, which also has the virtue of including the verse. Use the alternate changes. And actually, the Old RB chart isn't too bad. I wonder who at Berklee c.1974 was responsible for it?

Here's a chart with combined features, intended as a sort of common-practice-yet-correct version:

Finally, though, how you handle any song is your call, as long as it sounds good. This arrangement, for example:

Nov 7, 2016

Sax Reed Adjustment - Which Parts of the Reed Affect High Notes and Low Notes?

An important aspect of single-reed adjustment is where to sand or scrape in order to make certain notes speak more clearly. I've read quite a few books and articles on reed adjustment, in the never-ending quest for the elusive secrets of getting a good reed. Most writers seem to agree that lows are controlled by the lower end of the vamp closer to the shoulder, highs by the area closer to the tip. That is, if your low Bb is stuffy, then sanding or scraping towards the bottom of the vamp might help; if your high notes are not clear, try thinning towards the tip (but not too much). In my own experience, this is a valid concept.

A while ago, I was given a nearly complete set of Saxophone Journal magazines, dated from 1981 to 2001. I had (foolishly) never been a subscriber, but this was a great magazine, and now I have a second chance to read hundreds of informative articles. It will probably take me a year or two to check out every issue. Anyway, I just ran across an article (Nov./Dec. 1988) by mouthpiece expert Ralph Morgan, with a very interesting assertion - that if we divide a reed's blade into 33 equal segments and "each segment is weighed to compute its mass," each segment will be found to have a "resonance frequency" corresponding to a "quite accurate chromatic scale."

Morgan brings this idea up in a discussion of reed/mouthpiece compatibility, but obviously it would apply to reed adjustment as well. In his article, Morgan references research by Ignatius Gennusa, who was principal clarinetist with the Baltimore, Chicago, National, and NBC symphony orchestras, though it's not completely clear whether this idea comes from Gennusa. I admit to being a little skeptical about the precise validity of this claim - but I'd love to see the details of the research. On the other hand, I do agree with the basic concept that highs (and high overtones) are generated towards the tip, middle register notes in the middle of the vamp, and lows towards the bottom.

Perhaps you are familiar with this test for how to balance the left and right sides of a reed (quote is from this article on reed adjustment):
Turn the mouthpiece clockwise, so that your embouchure only controls the right side, with the left side of the reed free. Blow an open C# (sax) or G (clarinet) - then turn the mouthpiece the other way, to free the right side. If the “free” side (L or R) seems stuffy compared to the other, some wood should be removed from the stuffy side. 
Try combining this test with the idea that highs are controlled at the tip, lows at the shoulder - for example, if your middle D is stuffy, take some wood off approximately where the word "even" is in the diagram above, but first test that note with the mouthpiece-turning method to see which side is actually the problem; remove a little wood at the appropriate point on the reed for that frequency, and do it just on the appropriate (stuffy) side. It seemed to work for me.