Restoring the Original "Over the Rainbow"

Screen Shot 2019-10-06 at 12.14.32 PM (1).jpg

81 years ago tomorrow, on October 7th, 1938, Judy Garland and the MGM studio orchestra recorded “Over the Rainbow.” The song was written for the film by Harold Arlen and E.Y. Harburg, and arranged by Murray Cutter. It was recorded in the MGM scoring stage with its plywood walls, with Judy singing at the same time as the orchestra and Georgie Stoll conducting. They did eight takes that day, splicing together the beginning of take 5 with the rest of take 6 to be used in the film, and the rest is history.

Yes, it’s nice that it’s the 81st anniversary (and the 80th anniversary of the release of the film this autumn, too), but why bother to write this post? Well, because for a few weeks this summer, this particular arrangement of “Over the Rainbow,” which has loomed large in my life since I first heard it at age two, became a near-obsession. It all began on May 8th of this year, when in the middle of an email about another Garland arrangement I was restoring for an upcoming concert (more on that here), Michael Feinstein dropped the tantalizing tidbit that he might have found the original orchestral arrangement of “Over the Rainbow.” He couldn’t be sure until he received actual copies because he’d only gotten a brief glimpse of them among composer/bandleader/arranger David Rose’s files as he was helping David’s daughter, Angela Rose White, move some files from her Studio City office. 

And then we waited…

….and waited….

….and waited. 

So before continuing with the story, let’s just prolong the suspense and explain for anyone who doesn’t know the history of this arrangement why this is a BIG DEAL. James T. Aubrey — known as “The Smiling Cobra” — who took over as head of MGM in 1969 after bringing the world The Beverly Hillbillies, infamously trashed their film music library, and much of it was used as landfill under Interstate 405. Only the condensed 4-staff conductor scores of the films were kept, rather than all the sets of individual instrumental and vocal parts. The 4-staff score of “Production 1060,” The Wizard of Oz, did survive, but for “Over the Rainbow,” even the 4-staff score had gone missing. So every reconstructor of the Oz score has had to transcribe it entirely by ear from the film recording, and it’s a particularly hard orchestration to hear clearly, especially when instruments are covered by Judy’s vocal and they are all playing softly. So finding the 4-staff score and the instrumental part set would be like finding the Garland Holy Grail

Finally, on the sunny morning of June 7th, after a particularly rugged week of non-stop rain culminating in opening the kitchen cupboard to find water pouring out of it, I glanced at my inbox and saw an email from Michael Feinstein containing only the cryptic subject line “for your eyes only!!” and a zip file labeled “Over the Rainbow.” Without even stopping to make my jasmine tea, I dropped into my favorite chair with laptop and headphones to see whether this was, indeed, the original arrangement. The 4-staff score was obviously the missing one, or at least a copy, as it said “Property of Loew’s Incorporated” and “Prod. 1060” at the top. But looking at how many instruments were actually playing and WHAT they were playing, I wasn’t sure at first that these were the original parts. Moreover, the first violin parts I saw were dated May 25, 1940 and said “Property of David Broekman,” who was a violinist, conductor and arranger working in Hollywood at the time. Perhaps this was another, slightly later concert version? But as I listened more and more intently, part-by-part, it became clear, beyond a doubt, that what I was looking at was the original film version…and that there just might have been WAY more going on in that arrangement than I’d ever really heard before. 

At that point, if the guys on ladders cleaning our gutters happened to look in the bay window, they’d have seen me joyfully crying and jumping up and down in my pajamas. 

So, the plan quickly emerged that I would restore it and create a full conductor score and cleaned-up parts, and then Michael would re-premiere it as part of his MGM concert with the Pasadena Pops on September 14th. It was particularly meaningful to me to be entrusted with the restoration because it was the first song I can remembering falling in love with, and it was also the recording that, at the age of two, made me decide to become a singer….And, oh, yes, it could actually be the most iconic song recording of all time. So I knew I’d better do my level-best to get every note right.

But first, my inner Nancy Drew was curious to know how the parts and score had ended up among David Rose’s papers — thus felicitously saving the arrangement from becoming landfill. Get out your helmets and scuba cylinders because we are taking a deep dive! 

David Rose was, of course, Judy’s first husband and a prolific songwriter and arranger in Hollywood. The recopied-in-1940 violin parts that were initially cause for concern may have been used in a 1941 performance for a Greek Resistance benefit, for which no recording is known to exist. But the score and some parts (as well as the rest of the violin parts, which ARE original), are marked for the choral interlude that can be heard on the “Maxwell House Good News” preview radio broadcast of the Making of the Wizard of Oz on June 29th, 1939. That recording mostly corresponds with the parts, except that there are some obvious differences in the violins and that added choral interlude, written parts for which haven’t (yet!) been found. It’s my guess that David Rose used the score and parts to prepare the arrangements for that broadcast. They are also marked with a 2-bar introduction (a repeat of measures 1-2), which can be heard on the broadcast.

In another layer of markings, some of the trombone parts have “Thanks for the Memory (tacet) segué” and “Over Here” pencilled-in at the end, which led us to conclude that they were actually used on Bob Hope’s Pepsodent Radio Show on September 27th, 1939. Judy sang it on that broadcast, but no recordings are known to have survived. No recording, either, of the song during the Oscar ceremony of February 1940, when Judy won her special Oscar. 

Judging from clues in the markings, these parts may also have been used on September 24th, 1940 at the Golden Gate Exposition in San Francisco, when Judy sang the first part of the song accompanied by Harold Arlen himself at the piano, then the orchestra joins them on the last 8 bars of the first chorus (“Somewhere over the rainbow, bluebirds fly…”). By the time the next recording surfaces, from the Command Performance broadcast in 1943, it’s still the same key, but a different orchestral arrangement entirely. (Was it because they couldn’t find it??)

The MGM Studio Orchestra recording music for The Wizard of Oz. Note the plywood walls.

The MGM Studio Orchestra recording music for The Wizard of Oz. Note the plywood walls.

Finally, with a better idea of the journey these 26 instrumental parts may have taken, I settled down to painstakingly copy each of them and the vocal line into the music notation program Finale, and also make sure that each part corresponded with the film recording…because, of course, not every written note actually gets played in the final version that ends up onscreen. Things get changed or taceted “on the stand” all the time, and may or not be actually marked into the parts. 

Remember how I mentioned that it seemed like there was a lot more going on in that arrangement than one would have thought from a casual listen? Well, get ready for another deep dive here! (Just skip the next paragraph, though, if you’re running out of oxygen.) 

The biggest question mark hung over the violin parts. You’ll see in the first blurry measure of the conductor score pictured above that the violins are supposed to be playing sextuplets in four parts and that it supposedly continues — which they do, as it turns out. But they don’t appear in any previous transcriptions that I’ve seen or heard, and it took some very close listening to find them because they are playing very softly with mutes, and may not be close to a microphone. But if you listen closely you can faintly hear them playing the written sextuplets on the 2nd half of measure 1 and then beginning at measure 5 onwards. And in the subsequent unused faster takes from the session (on the Oz DVD release and here’s one), the violins are more often audible throughout — when they had to play it faster, they got louder. Moreover, there is a subtle, impressionistic feel of movement and a wash of sound in that frequency range that can’t be accounted for anywhere else in the orchestration. All of this led to the conclusion that they were actually playing the parts as written throughout, and the conductor was just sitting on them to keep the volume down. Additionally, the parts are fairly marked-up, but there’s no indication at all of anything being cut. Later, I ran across a quote from Murray Cutter (told to Aljean Harmetz, who wrote The Making of the Wizard of Oz) about how he approached the song that he would forever be known for: He said his arrangement was “as pretty as I could make it, lots of strings and a touch of woodwind.” 

You can also hear the violin sextuplet figures very distinctly on the short introduction into the song, which was actually written and recorded months after “Over the Rainbow” and clearly made to fit with Cutter’s orchestration. 

An interesting side note: In the first violin part of Judy’s early-‘50s concert arrangement — the one she also sang at Carnegie Hall in 1961 — arranger Hal Mooney exactly duplicates the Cutter arrangement for the first two A sections, up through the beginning of measure 15, in an obvious homage to the earlier version. By this time Judy and David Rose were long divorced, but Mooney would likely have had access to the MGM music library, which still existed at that time and may have had copies of the parts, even if the 4-staff score was hiding in David Rose’s files.

So the existence of those shimmering violin parts is a huge missing piece restored. Another major difference is that the “happy little bluebirds” melody in the orchestral interlude is actually being played by three muted trumpets in unison, not by the flutes indicated in previous transcriptions. Cutter originally scored it for trumpets in three parts and up the octave, and it must have been changed on the stand sometime during takes 1-4 and no one bothered to pencil-in the change…so that when they used the parts months later on the “Good News” broadcast they played it as written. 

These revelations and many others (see footnotes below) meant that it was likely that no one had really heard the full glory of Murray Cutter’s original arrangement since that first recording session on October 7th, 1938 — until last month. I wish I could have been a little bluebird on September 14th, when Michael Feinstein and the Pasadena Pops, with Broadway star Karen Ziemba, raised the scrim on the arrangement. Rumor is, it sounded beautiful. But, alas, I will have to wait a little longer to hear it. Fortunately, it is on the program January 25th for my symphonic Garland concert, “Get Happy!,” with the Ashland Symphony and conductor Michael Berkowitz, and this time I’ll get to sing it. Since I think I’ve probably been subconsciously waiting for this moment for the past 46 years, what’s another few months?

Many, many thanks to Michael Feinstein for the undreamt-of chance to help preserve this song that goes so deep into our musical consciousness, as well as for letting me share in the fun of the treasure hunt, to Oz-and-Garland-expert John Fricke for his historical sleuthing and sharing in the enjoyment of the chase, as well as to the ever-on-top-of-things Starleigh Goltry, who took detailed additional photos of every page and tracked down Jim Hardy, the Bob Hope archivist, in an attempt to find a recording of the Pepsodent Radio Show in question. Any factual errors, however, are entirely my own! The last and biggest thank you has to go to David Rose for NOT returning that file.


Those few other mysteries if you’re really curious: 

1.) I kept swearing I heard a bass clarinet playing a portamento-ing bass line near the end of the song, but there was no bass clarinet part in the part set Michael found. So I went back and checked the instrumental break-down for the session and sure enough, there were FOUR clarinets that day, even though we only had parts for three. So I reconstructed that part from what I could actually hear and what might have been likely when I couldn’t actually hear it.

2.) We know from the instrumental break-down that a guitar was also there, but no guitar part survived and no guitar is actually audible on the recording except briefly between takes. It’s possible that it’s only playing when the harp also plays, or playing an inaudible rhythm part, but we may never know for sure.

3.) There was a drum set part marked “Brushes if wanted” and pianississimo, and a drummer at the session, but it’s also inaudible and probably just got cut at some point. Maybe the drummer played the bird whistle, instead…

4.) The harp part is rather different than written in some key spots, with some pencilled-in markings supporting what is being played, but not exactly, so some transcription was necessary there. 

5.) There was a tuba part in the set that was clearly added for a later performance, as it includes a 2-bar introduction and is in a different hand without the stamped “Over the Rainbow” title that all the other original parts bear. 

About the recorded takes:

The first four takes have been lost, but takes 5-8 survived and can be heard on the recent Oz DVD release and elsewhere. Takes 5 and 6 were the ones used in the film; Judy coughs at the end of the first section of the song, so they used the take up until the cough and then spliced it with the rest of take 6. Then they did two faster-tempo takes, with a short bird whistle only at the end of take 7, but not on the interlude as in take 8. (As a side note, take 6 is slated mistakenly as take 7 in the session, and then the actual take 7 isn’t slated at all, but thankfully we have the session logs and it’s all there accurately.)