The Bridge to “Ipanema”

There's been some buzz lately around a video from YouTuber Adam Neely in which Adam discusses Jobim's "Girl From Ipanema" ("Garota de Ipanema"), including his explanation of Jobim's harmony on the bridge. Several musician friends have sent me the link, and there have been two threads on saxontheweb.com discussing the video (here and here). Last time I looked, this video had 1,226,782 views.

I've enjoyed other YouTube videos from Neely, but this one has some questionable information, IMHO. I'm going to take the bait and write up my comments. Here's his "Ipanema" video, if you haven't seen it yet:

What is the proper key for this song?

The first issue that jumps out at me is the question of the proper key for "Ipanema." All of the published sheet music versions, and all of the fakebooks that I have seen, show the song in F. Adam tells us that this key is a "relic of American cultural hegemony, codified by decisions made at the Berklee College of Music in the 1970s." Strong words! He is referring to the Real Book (1974), which shows the tune in F.

Neely states that in Brazil, F is not regarded as the hip key for this song, and that for a Brazilian audience, you had better play it in Db, because "Db is thought of as the more authentic, Brazilian key." Why, you might ask? Because Db was the key used on the "Getz/Gilberto" album, the recording that made the song a worldwide hit in 1964.

I’m certain that Jobim always intended for this tune to be in F. There are several original Jobim manuscripts that you can view at jobim.org, all of them in F. Whenever Jobim performed it, it was in F, as nearly as I can tell from his live videos, except when he played it with Joao Gilberto. I played with Brazilian bands in the SF Bay Area for years, and the key for "Ipanema" was always F; no one ever suggested otherwise. My musician friends tell me the same thing. I have never heard that any other key is regarded as more authentic. But perhaps Neely has experience I don't have, or different sources of information.

The simple explanation for recordings in other keys is that those keys were chosen to accommodate the singer's best range. That's done all the time. I think it was in Db on the album because Joao Gilberto preferred that key. On some later recordings, Joao plays it in D.

Here is a 1962 live version with Jobim, Gilberto, and Vinicius de Moraes. They play it in F:

This reminds me of a story. Years ago, I was on a big-band gig for a corporate event. Eddy Arnold, the country singer, was at the event for some reason, and was scheduled to sing "September Song," accompanied by our pianist, Reed Struppa. Before the gig, when we were setting up, Eddy asked Reed to find the key where he would be most comfortable. They rehearsed a little, and came up with some little-used, awkward key. After the gig, Reed told me, "When they do that, I tell them OK, then when the time comes I just play it in the nearest easy key, C or F or whatever. They never notice the difference."

Harmony of the A section

Anyway, back to Ipanema. Adam gives us some basic bossa nova history, and points out that the A section of Ipanema uses essentially the same harmony as the A section of "Take the A Train" (that is, I for 2 bars, V of V for 2, then II V I, with a bII tritone sub for the V in "Ipanema"). Here I agree with him. Incidentally, this harmony was not original with Billy Strayhorn, who wrote "A Train" in 1939. Strayhorn borrowed it from Jimmy McHugh's 1930 tune "Exactly Like You." Neely notes that this progression also appears in the A section of Jobim's "So Danco Samba." I'd add that Jobim's "Desafinado" follows this progression too, for the first 6 bars, before veering off into creative territory.

Here’s a video of Jobim playing “Ipanema” in Sao Paolo in 1994 at an “All Star Tribute.” It's a fairly good band: Jobim, Herbie Hancock, Joe Henderson, Ron Carter, Shirley Horn, Gal Costa, Jon Hendricks, Gonzalo Rubalcaba, Oscar Castro-Neves, Paolo Jobim, Harvey Mason, Alex Acuna. They play it in F. At 55:39 Jobim sings the "A Train" melody over the Ipanema A section, a sort of inside joke.

The bridge

Neely points out some countermelody lines in the bridge (13:27 in his video) that Jobim clearly intended to be part of the song. Adam criticizes the Real Book compilers for not including these countermelodies. I agree that it would have made for a better chart, but let’s not be too judgmental; the Real Book was a great product for 1974. By the way, the current Hal Leonard legal version of the RB doesn't include the countermelodies either.

When Adam gets into explaining the harmony of the bridge to Ipanema, he gets into questionable territory, as I see it. Here is the bridge:

At first glance the chords do not seem to follow the "rules" of standard jazz harmony. But it makes sense to me this way:

Gbmaj7: New key, up a half step from the A section (think of it as F#maj7). 

B7: IVb7 in F#, a blues-type IV chord. Introduces the blue note A natural, the b3 in F#. 

F#m7: It's really A6 with F# in the bass, the I of a new key. 

D7: IVb7 in A, blues IV chord again. Introduces blue note C natural, b3 in A. 

Gm7: It's really Bb6 with G in the bass, the I of a new key. 

Eb7: Again, it's IVb7, a blues IV chord. Introduces the note Db, b3 in Bb. 

The last 4 bars are a turnaround back into F, to set up the last A section.

The bass notes on the F#m7 (bridge bar 5-6) and D7 (bar 7-8) have an interesting effect and a certain logic. F# has continuity with the first tonality in the bridge. D7, besides being IVb7 in A, is V of the Gm7 that follows. These features do not interfere with the key centers Gb - A - Bb as described above.

As Neely remarks, there is a virtual "cottage industry" of YouTube videos and articles that try to explain the bridge. If you want to get into it, do a YouTube search for "Ipanema bridge analysis." Don't say I didn't warn you.

Neely's take on the bridge starts at 10:12 on his video. He hears the Gb as a IV in Db, B7 as a bVIIdom in Db, F#m7 as a II in E, D7 as a bVIIdom in E; Gm7 as a II in F, Eb7 as a bVIIdom in F.

It's an interesting take. With some effort, I can hear it as he describes. I can't say he is exactly "wrong," but his explanation doesn't really work for me.

I don't want to get too deep in the weeds here, but:

1) I don't hear the bridge as starting in the key of Db (that would make Gbmaj7 the IV, with a Lydian tonality). I hear it in Gb, starting on the I. Jobim very often used major 7th melody notes over tonic major 7 chords, as he does in the A section of this song, and that's what we have here. 

2) I'm not the only one who hears it that way; some other YouTubers in this "cottage industry" agree with me, as do some of the commenters in the saxontheweb threads on this subject. 

3) Occam's Razor favors my explanation.

Neely goes on to say that he hears the countermelody as a blues lick. I don't. It's true that you can shoehorn the notes of the countermelody lick into a blues scale that would seem to fit Neely's conception of the key centers, but I just don't hear it.

Of course, reasonable people can disagree.

Here's yet another take on the bridge, more scales-based, from my friend Larry Lewicki:

FWIW, I  don’t analyze the bridge the same way as Peter. I just see scales with shared pitches. I almost picture Jobim doing his scales F# major, F# melodic minor, F# dorian ... just changing one pitch. 

F#Δ  - F# major

B7#11 - F# melodic minor (an essential element of blues - the IV7) (shares 6 notes with F# major) 

F#-7 - F# dorian (shares 6 notes with F# melodic minor) 

D7#11 - A melodic minor - shares 5 notes with F# dorian (check Tenderly bars 5-6, 21-22)—- D7 is the dominant of G minor 

G-7 - G dorian

Eb7#11 - Ab melodic minor - shares 5 notes with G dorian (Tenderly cadence)—- tritone sub of Eb7 is A7 - Am7 is very close - that’s the beginning of the final 2 bars - Bebop iii VI7 ii V7 cadence in F. 

I don’t see the F#-7 as a inversion for an AΔ.... because of the dorian mode D#

Any of these analyses - Neely's, mine, Larry's, and the various YouTubers' - can serve as the conceptual basis for a perfectly good solo, depending on how the player's melodic sense operates. In that respect, they are all just fine. However, if I'm trying to figure out what Jobim may have been thinking, I kind of like my approach.

Harmonic ambiguity as a defining feature of bossa nova

Adam states that the harmonic ambiguity he perceives in "Ipanema" is typical of bossa nova. I think that as a generalization, that's somewhat of an overstatement. Mostly, the classic bossas can be explained fairly neatly in standard jazz terms.

Jobim was a particularly creative composer, a generation younger than the "Great American Songbook" composers, building on their work. And even in Jobim's songs, especially in the earlier ones that made him famous, most of the harmony is pretty straightforward.

Does "Ipanema" have an introductory verse?

Quite a few Great American Songbook standards had an introductory section called a "verse." At 30:28 in his video, Adam points out the introductory verse in the live 1962 Jobim/Gilberto/Vinicius recording. I hadn't known about that. Desafinado has a verse too, that Jobim used in performance. He even had English lyrics for it. But the "Ipanema" verse was news to me. Thanks to Adam for pointing it out!

It does seem that this verse was not exactly intended to be part of the composition, though, but rather was created for a particular occasion.

I asked my friend Guto for a translation, and here it is, with his comments:

That intro is indeed very interesting. It seems it was a one off for that album, almost like an insider joke the trio Tom, Joao, Vinicius was telling to the live audience before the song starts. They go: 

João: Tom, e se você fizesse agora uma canção?

 Tom, how about you now make a song?

Que possa nos dizer, contar o que é o amor  

One that says, tells us what love is 

Tom: Olha Joãozinho, eu nao saberia

 Look dear João (or little Joao, in an affective way), I wouldn’t know how

Sem Vinícius pra fazer a poesia  

Without Vinicius to make the poetry 

Vinícius: Para essa canção se realizar  

For this song to come together 

Quem dera o João para cantar

 I wish João would sing

João: A, mas quem sou eu, eu sou mais vocês  

Oh, but who am I, I’m more you both (in the sense of I trust you’d do a better job) 

Melhor se nós cantássemos os três

We’d better sing the three of us

Guto also noticed some musical jokes in the verse:

... the second line Tom sings, “sem Vinicius pra fazer a poesia,” sounds the same as this section from Desafinado:

and the last phrase from Joao, “melhor se nos cantassemos os tres,” sounds like the ending phrase of One Note Samba. 

Besides the 1962 live recording, the only other place I've heard this verse is in this great video with Jobim and Gilberto, I think from 1992. Joao sings the verse himself, and they play the tune in D:

Pery Ribeiro’s 1962 version

Beginning at 23:12 in his video, Adam discusses a 1962 version by singer Pery Ribeiro that may predate even the 1962 Jobim/Gilberto/Vinicius recording. Because it is probably the earliest recorded version of the song that we have, Neely finds importance in 1) the key it's played in, and 2) the harmony used for the bridge.

This version is in G, leading him to say that "The original is in the key of G." But there's no evidence that Ribeiro's was the "original" version. This is an unwarranted assumption. More likely, the key was chosen to accommodate Ribeiro's voice.

About the harmony in Ribeiro's version - Neely presents this arrangement as a sort of "missing link" between a hypothetical Tin Pan Alley harmonization and the "bossa nova" final version. But it's quite likely that the harmony used in this recording was the work of Ribeiro's arranger; it was not necessarily an early Jobim version of the song.

Neely seems to be saying that the "final" version of the bridge harmony was actually the work of Joao Gilberto, editing and simplifying Jobim's "original" harmony as used in the Ribeiro recording. But there are two unwarranted assumptions here: 1) that the Ribeiro harmony was Jobim's, and 2) that it was Gilberto who created the "final" version.

I really did enjoy Neely’s video, in spite of a few disagreements. Hopefully some of those million-plus views will get some younger musicians interested in Jobim and bossa.

To close, if you'll forgive me, here is an old musician's joke, that an old musician told me during a band break:

A jazz group has a gig at a bar in Chicago. The bass player lives across the border in Indiana. He has a history of showing up late for gigs. When it's time for the downbeat at 9:00, the bass player still hasn't shown up. At the first break at 9:45, still no bass player. The band leader is getting increasingly angry. Finally, at 10:30 the bartender goes up to the band leader and says, "Your bass player is on the phone. He's stuck on the bridge to Indiana." 

The bandleader says, "Man...there is no bridge to Indiana."

The "Groovin' High" family of II V's

In a number of classic bebop melodies, and in classic recorded solos as well, you'll hear examples of what I'll call the "Groovin' High" family of II V licks. The common feature of these licks is the use of a chromatic line from the root of the II chord, descending by half steps, to the third of the V chord. 

Here's an example from Dizzy Gillespie's 1945 composition "Groovin' High" (transposed here to the key of C). This II V phrase incorporates and embellishes the chromatic line D, C#, C, B, as indicated by the arrows:

Here's another example from the head to Charlie Parker's "Billie's Bounce," also from 1945 (transposed to C). Again, note the line D, C#, C, B:

Another well-known tune that uses this line is Sonny Rollins' 1956 blues "Tenor Madness" (transposed to C):

To understand where this chromatic line comes from, consider the example below. The "active ingredient" in any II V is the motion of the seventh of the II chord (C) dropping to become the third of the V chord (B). No other notes need to change when changing chords. The notes D and F are common to both Dm7 and G7, and the note A can stay, to become the ninth of the G7.

basic voice leading for II V in C

In the G7, B is the most important note, as it is the leading tone that is providing tension, trying to resolve up a half step to the root of the I chord (C, C6, Cmaj7, etc.) that normally would follow.

The example below shows an elaboration of this voice leading, adding a chromatic line starting on D. The line implies the chords shown here, though the melodic line does not really need to have the support of these exact chords.

with descending passing tones added

This voice-leading line can be further elaborated to create II V patterns like the ones below. Chords might be shown (or played) as in the example above, or as in the first example below, but more often are just squared off to one per measure, as in the other two examples:

Charlie Parker was unquestionably aware of this II V device. Below are some excerpts from melodies and solos in the Charlie Parker Omnibook, Vol. 1 (Eb edition). Coincidentally, these are all in the key of G:

from "Confirmation" head (the D7 notes use a
#9 b9 (F and Eb) rather than a normal third (F#)

from "Anthropology" head (I'd have shown the first measure as Am7)

from "Anthropology" solo (likewise)

from "Bloomdido" solo (Charlie is displacing the implied harmony a bit)

from "Blues (Fast)"

I chose to call this device the "Groovin' High" family of II V's because that tune and "Billie's Bounce" are the earliest jazz examples I could find, both from 1945. I have no doubt that jazz pianists had figured this out earlier, and I'm sure that plenty of similar melodic and harmonic examples can be found in the classical repertoire. One possible earlier jazz use could be Thelonious Monk's "Round Midnight," in the chord changes at the beginning of the A section. Monk may have written this tune in the 1930s. But there's no way of knowing if he was harmonizing it the same way then as he did in later recordings. 

If any readers know of earlier examples in the jazz repertoire, or in Great American Songbook tunes of the 20s -30s, please leave a comment!

(Erratum: In the example labelled “with descending passing tones added,” the first chord symbol should of course be Dm, not D)

"There Will Never Be Another You" - Why does F7 resolve to Eb major?

In older lead sheets for some jazz standards, you will sometimes see II dominant chords (that is, V of V) progressing directly to the tonic. In the key of C, that would be D7 moving to C major. This sequence is found in "There Will Never Be Another You," "It Might As Well Be Spring," "Memories of You," and "Time After Time," among others.

In newer charts for these songs, this progression is often replaced with some sort of reharmonization "workaround" that better fits current preconceptions of how harmony should function.

D7 resolving to C might seem to somehow transgress the rules of harmony, but if a composer wrote this chord change, there was probably a good reason. In the original sheet music for these songs, the piano arrangements will usually show the II dominant going to a tonic chord that has its fifth in the bass. In the key of C, that would be D7 going to C/G (C major with G in the bass) 

This is a classical device, the "cadential six-four chord." In classical terminology, a triad with its fifth in the bass (2nd inversion) is called a "six-four" chord. Although C/G might appear to be a tonic chord, it is actually functioning as a V with two suspended notes. In jazz terms, C/G could be regarded as a sort of Gsus4 chord. Generally, it will then go on to a typical V  I final cadence. If you look at it that way, the D7 to C/G progression is not so odd after all.

In the original sheet music for songs using this harmonic device, the piano arrangement may include a fifth in the bass of the tonic chord, while the chord symbols may fail to mention it. If a lead sheet is created from these chord symbols, players will just see D7 going to C, and may feel that it is flawed writing, that should be fixed - hence the reharmonizations in many modern lead sheets.

Of course, a reharmonization may have been made because it just sounds better to the arranger, or because it offers better opportunities for improvisation. The composer's original harmony may not be what is best for you. Still, it's interesting to see why these songs differ in various printed sources, and what replacements musicians have come up in an effort to avoid the V of V to tonic six-four device.

There Will Never Be Another You

Harry Warren's "There Will Never Be Another You" provides a great illustration of how this device has been avoided and rewritten in contemporary fakebooks. Here's the last page of the original sheet music. Click to enlarge.

"There Will Never Be Another You" is in Eb. The V of V to tonic six-four (F7 to Eb/Bb) occurs in the first two measures of the last line (mm 28-29 of the song, not counting the introductory "verse"). 

Although the chord symbols show Ebdim over the word "If" on beat 4, IMO this chord (it's actually a 4-note Ebdim7, as there's a C in the melody) is functionally insignificant. The Ebdim chord symbol is only shown to make a guitarist's comping fit with the F# note in the piano part; that F# is just there to provide a passing harmony with the C melody note on beat 4. 

The F7 (V of V) provides a strong push towards a V (Bb7), but instead of a V, the next measure has a cadential six-four (Eb/Bb, though the chord symbol shows only Eb). This is a textbook usage of the cadential tonic six-four. It makes me think that perhaps Harry Warren (Salvatore Antonio Guaragna) had some classical training.

Although the chord symbols show F7 for beat 3 of bar 28, there is no F note in the piano part. Nevertheless, I agree that the functional chord is actually F7; the bass line moving from Bb to A is typical of a II V (Cm7 to F7). 

Apparently many jazz players have found the indicated F7 to Eb sequence to be unsatisfactory, as it has been reharmonized in a number of different ways. 

Here are the last 8 bars of the tune as shown in various fake books:

Old ("Classic") Real Book:

Although this is the first version I learned, the Gm7  C7 never sounded right to me.

Hal Leonard "6th Edition Real Book":

The Am7 D7 here sounds better than the Gm7 C7 in the old RB version. Comparing this version to the original piano part, you might convince yourself that the D7 corresponds to the original F7 Ebdim7, in the sense that D7b9 is kind of equivalent to Ebdim7, and Ebdim7 is kind of equivalent to F7b9. However, IMO, this has nothing to do with Warren's original intended progression.

"New Real Book" (Sher Music): 

This is the same as the "6th Edition" version above, but with b5 added to the Am7. To their credit, the editors showed F13 as an alternate chord change, though without showing the following Eb/Bb.

Hal Leonard Real Jazz Standards Fake Book: (This useful book shows slightly edited sheet music chord symbols above, with a more modern reharmonization below):

In the "revised" changes, F#dim7 occurs on beat 3 rather than on the original sheet music's beat 4 (except for the root, F#dim7 and Ebdim7 are the same set of notes). I see the function of the F#dim7 here as being different from the Ebdim7 in the original sheet - in this chart, it's there not so much to accompany a piano arrangement that has an F# note below the melody note C, but rather is there to put passing tones between the F7 and the Ebmaj7 so that the change does not seem so abrupt. 

"Colorado Cookbook":

Here we see the same Am7 D7 as in the "6th Edition Real Book," but resolving to Gm7 rather than Eb. Perhaps this seemed like a good solution because D7 is V of Gm7, and Gm7 is a reasonable substitute for Eb major.

"The Book":

Here we see Cm7  Ebdim7 going to Eb. It's like the chord symbols in the original sheet music, but leaving out the F7 that should be the main functional element here. This doesn't make much sense. It misses the original intent, and results in a weak bass line.

Hal Leonard "Ultimate Jazz Fake Book":

(Click to enlarge.) This version fills bar 28 entirely with an F#dim7 chord. This chord does not follow logically from the Cm7 that precedes it, and does not make for a good bass line. The only logic here is that (1.) diminished chords can pretty much pivot from anywhere to anywhere, and (2.) F#dim7 is sort of like the F7 in the original sheet music, though of course the original V of V to six-four intent is lost.

All in all, it seems that jazz musicians have gone to considerable effort to avoid and rewrite the V of V to cadential six-four device in Warren's tune. 

Below is a list of some more tunes that use the V of V to cadential six-four, with some sources where you can see the progression (thanks to Tom Simpson for this list). I only had the original sheet music for a few of these. 

Embraceable You (mm 12-13, 28-29) (see Sher "Standards Real Book") 

I Remember You (mm 28-29) ("Real Jazz Standards Fake Book")

I've Got A Crush On You (mm 14-15) (Old "Classic Real Book vol. 3")

It Might As Well Be Spring (mm 36-37, not counting intro "verse") ("Real Jazz Standards Fake Book")

Lover Come Back To Me (mm 4-5) ("Real Jazz Standards Fake Book")

Memories Of You (mm 28-29) (Old "Classic Real Book vol. 2")

Nice Work If You Can Get It (mm 4-5, 28-29) (Sher "Standards Real Book")

I Surrender Dear (mm 4-5) (Django Fake Book)

Time After Time (mm 28-29) (Sher "Standards Real Book")

My Romance (mm 28-29) ("Real Jazz Standards Fake Book")

It's interesting to note how many of these instances occur in measures 28-29, as part of the final cadence.

If you can think of any more tunes that use this device, please leave a comment!

The "Honeysuckle Rose" II V lick - Charlie Parker, K-Ximbinho, and Mauricio Einhorn

One of the staple bop solo clichés is the "Honeysuckle Rose" II V lick. Bop-era jazz players utilized it quite a bit, and it is still part of the jazz improviser's language.

Here's the lick as it originally appeared in Fats Waller's 1929 song "Honeysuckle Rose" (key sig. is one flat):

Use of the lick in soloing is associated with Charlie Parker, but I really can't say when American jazz players first started using "Honeysuckle Rose" as an improv element. If any reader can point out a recorded 1930s instance in a solo by a jazz artist, please let me know in the comments section below.

One interesting early use is in the head to Tadd Dameron's "Lady Bird." According to one source, this tune was written around 1939. At that time Dameron would still have been living in Cleveland, Parker was not well-known, and bop had not yet taken shape as a distinct style. (See this post for more about this tune.)

Here's the "Honeysuckle Rose" phrase in "Lady Bird" as it appears in the Hal Leonard Real Book (bars 3-4). The last note was originally F, not E.

Another well-known use of the lick is in "Donna Lee," written by either Miles Davis or Charlie Parker in 1947 (see this post regarding authorship) (key sig. is 4 flats)

By Parker's own account, "Honeysuckle Rose" was the first complete tune that he ever learned, as a teenager. He used the lick throughout his career.

Parker got extra mileage out of the "Honeysuckle Rose" lick by varying its notes and rhythmic placement. A few examples from the Charlie Parker Omnibook are below:

Marmaduke head (Omnibook, p. 68)

Scrapple from the Apple solo (Omnibook p.17)

Blues for Alice solo (Omnibook, p.19)

Bloomdido head (Omnibook, p. 108)

7

Donna Lee solo (Omnibook, p. 49)

Perhaps solo (Omnibook, p. 72)

Klaunstance solo (Omnibook, p. 90)

Parker uses the lick in various ways, not just over a straight II V. For example, the "Donna Lee" solo clip above is over what I would have called IVm bVIIdom (as is the Dameron "Lady Bird" instance). The Scrapple example is pretty heavily disguised, and played in Eb, "side-slipping" a half step above what the rhythm section is playing. 

I also ran across a couple of examples in some Brazilian classics - a 1946 choro by K-Ximbinho, "Sonoroso," and a 1960's bossa by Mauricio Einhorn, "Estamos Ai." That seemed a little surprising, but upon further reflection, maybe not so surprising after all. K-Ximbinho (stage name of Sebastião Barros) was a Brazilian jazz saxophonist and clarinetist, born in 1917. His bio is here (translation from Portuguese Wikipedia). It's quite likely that he would have been listening to records from the US in 1946.

Mauricio Einhorn is a jazz harmonica player, born in 1932. He was active as a performer in Rio in the 1950s, and was prominent in the early days of bossa nova. He's still performing. American jazz fans might remember him as the composer of "Batida Diferente," recorded by Cannonball Adderley. Einhorn certainly would have been familiar with the bebop language. His bio is here (translation from German Wikipedia).

Here's the lick as used in K-Ximbinho's "Sonoroso" (it occurs at the end of the "B" section) (key sig. is one flat):

There are many recordings of  "Sonoroso" on Youtube. The first one below is probably the original. It's from 1946 with the Orquestra Tabajara; K-Ximbinho played with the group at that time. Following it is a more modern version with Altamiro Carrilho and Paulo Sergio Santos.

Here's the "Honeysuckle Rose" lick as it occurs in Einhorn's "Estamos Ai" (bars 16 and 32):

A 1965 version by Leny Andrade:

A two-harmonica rendition by Einhorn and Toots Thielemans, played with a swing beat:

"My Romance" - What are the right changes?

Rodgers and Hart's song "My Romance" first appeared in the musical "Jumbo" (1935), produced by impresario Billy Rose. From Wikipedia:

The musical opened on Broadway at the Hippodrome Theatre on November 16, 1935 and closed on April 18, 1936 after 233 performances. Directed by John Murray Anderson and George Abbott, it starred Jimmy Durante, Donald Novis, Gloria Grafton, and a number of circus specialty acts. Jumbo tells the story of a financially strapped circus. At the end of each performance, Durante lay down on the stage and permitted a live elephant to place its foot upon his head.

The large 5,000-seat theatre was turned into a circus tent where the various specialty acts (including acrobats and animal acts) performed during the show. The music was played by Paul Whiteman and his orchestra.

Other notable songs that first appeared in this musical were "Little Girl Blue," and "The Most Beautiful Girl in the World."

Incidentally, this was not the first show at the Hippodrome that had featured a live elephant. The song "Poor Butterfly" (1916) was written in an elephant pen in the basement of the Hippodrome; Harry Houdini performed a "disappearing elephant" trick there too.

Here's a newsreel from 1935, covering the opening of "Jumbo," including some rehearsal footage:

In a recent interview, bassist Steve Swallow mentions that the chords to "My Romance," as shown in the old bootleg Real Book, are the "Bill Evans changes." This, of course, sparked my curiosity about what the original changes were. 

I was able to locate a copy of the original sheet music. In looking at "Golden Age" standards, original sheet music is generally a pretty good indicator of the composer's intention, as it is likely that the composer approved it for publication. Besides that source, I have a version of the song from the "Tune-Dex" fakebook, a popular bootleg from (I think) the early 1950s, which shows bare-bones chord changes for hundreds of standards, probably copied from the sheet music, with a high degree of inaccuracy. 

We can also refer to this 1936 recording by Donald Novis and Gloria Grafton, the stars of "Jumbo." There is a good chance that this arrangement is at least somewhat close to the way the song was presented in the show:

There are two modern fakebooks that show some version of the original sheet music changes, side-by-side with a modern reharmonization - The Best Chord Changes for the Most Requested Standardsby Frank Mantooth, and the Real Jazz Standards Fake Book, both published by Hal Leonard.

The sheet music puts the tune in the key of C, as does the Tune-Dex book, but for some reason the 1974 bootleg Real Book shows it in Bb (the Hal Leonard "6th Edition" Real Book, copying the old RB, also has it in Bb). The 1936 recording is in Eb, probably to accommodate the singers. Bill Evans played it in C. To make comparisons easier, the charts in this post are all in the key of C.

The sheet music includes a "verse" (introductory lead-in) that no one really uses any more; I won't discuss it here. It's included in the "Real Jazz Standards" version.  

In deference to copyright, I won't post the complete sheet music, but I think it's within "fair use" to show this clip:

As is usual in old sheet music, the chord symbols do not really represent the piano arrangement. Below is a chart with the chord symbols as shown on the sheet music:

However, the actual piano notes indicate a harmony closer to this:

Some features in the piano arrangement: 

1) stepwise bass lines in mm1-2, 5-6, 17-18, 21-22. The chord symbols miss this. Modern charts suggest different bass lines here (see "Evans" chart, below).

2) triads with added 9 - a sweet, bright sound.

3) a lot of IV chords - sweet and simple, in keeping with the theme of the lyrics.

4) F#m7 rather than F#m7b5 in m13. Modern charts all use F#m7b5, presumably because it seems like the right way to do a minor-key II V. That's not what Rodgers wrote, though.

5) D9 in m28 resolves to C/G in m29. In classical terminology, C/G is a cadential tonic six-four chord. This device (II dominant to I with fifth in the bass, followed by a final cadence) is common in classical music. It occurs in the original harmonic setting of several jazz standards, e.g. "After You've Gone," m16; "I Remember You," m32; "There Will Never Be Another You," m28. The fifth in the bass sets up a brief dominant pedal for the cadence.

The II dominant to tonic six-four device has fallen out of favor in jazz. It seems like just about every modern chart of the tunes listed above replaces the IIdom approach with something else. Some modern charts of "My Romance" preserve the G in the bass under the tonic C chord, keeping the dominant pedal effect, while replacing the D7.

Here's an image of the beginning of the Tune-Dex chart:

This chart really cuts to the essence of the tune. The Tune-Dex fakebook was probably the most complete bootleg collection of tunes available to popular music performers in the late 1940s and early 1950s. I got my copy from the father of one of my students; his father ("Doctor K.") had been an aspiring musician in New York in the early 1950s. The notations in the margin are Doctor K.'s; he was studying the Schillinger system. The Schillinger symbol ΣIV indicates a #4 melody note; it occurs in m3. The note in the margin says, "going to ΣIV more surprising more intense."

For comparison, here is a chord chart with the bare-bones Tune-dex changes. Try playing the melody with this harmony - I think it preserves the simple, sweet quality of the tune pretty well: 

As Steve Swallow relates in his interview, the old bootleg Real Book was created in 1974 by two Berklee students. Many of their charts originated with some of their instructors, or friends of their instructors. Swallow describes the "My Romance" chart as using "Bill Evans" changes. Here are the chords as shown in the old RB, but transposed to C, rather than the RB key of Bb:

Some features of this version:

1) The bass line in mm1-2 and 17-18 is changed to a familiar I  II III bIIIdim progression. The bass line in mm 5-6 and 21-22 is replaced with a familiar "My Funny Valentine"-type voice-leading line over a static minor chord - the line here is A to G# to G, over the A minor chord.

2) The measure-long IV chords are gone.

3) In mm9 and 11 a bVIIdom chord (Bb7) is added to transition from IV to I - again, a standard device. 

4) In m14 Eb7 is added, a tritone sub approach to the Dm7 that follows.

5) mm27-28 are reworked, replacing a short chain of dominants with a minor II V and a II bIIdom. It's a fancier way to get there, but the Ab7 still sets up a tonic six-four chord.

I think you will find that virtually all modern charts derive from the old RB "Bill Evans" changes, with a few alterations here and there. If you want to learn the tune today, you should probably learn this version. Personally, I favor the key of C rather than Bb. The New Real Book vol. 1 (Sher Music Co.) has the tune in C, with Evans-like changes.

I don't know exactly where the old RB compilers got their chart, but it's pretty close to the way Bill Evans plays "My Romance" in the 1961 Village Vanguard recording:

Here's a link to a transcription by Jorn Swart of Bill's first chorus.

Bill Evans played "My Romance" through his entire career. He recorded it a number of times, including on his first album, "New Jazz Conceptions" (1956) and his last, "The Last Waltz" (1980). Of course, the lead sheet does not even begin to show what he did with the tune; his approach evolved continuously. Here's his last recording of "My Romance":