Skip to content

Horenstein and Mahler’s Third Symphony Revisited

Horenstein and Mahler’s Third Symphony Revisited

By John H. Haley

Harmony Restorations, LLC

Many music lovers treasure Jascha Horenstein’s classic 1970 LP recording of Mahler’s Symphony No. 3 with the London Symphony Orchestra on the Unicorn label, released on LP in the U.S. on the Nonesuch label (remember that psychedelic cover?). It was recorded in six sessions that occurred on July 27, 28 and 29 at the Fairfield Concert Hall in the London suburb of Croydon, by well known audio engineer Bob Auger and his team, Harold Lawrence producing. The Unicorn recording was multi-miked, recorded in eight tracks on one inch tape on Scully recording equipment. 

A little-known fact is that a simultaneous recording using totally different equipment operated in a separate control room was made concurrently at the same recording sessions by a leading American recording engineer, Jerry Bruck, who was invited to do so by the general manager of Unicorn Records. Except for a borrowed four-track Scully recorder and Dolby A unit, Bruck brought his own equipment with him from the U.S., including microphones, using a unique mic set-up to achieve a spectacular four channel recording that has never been released or even heard before by the public. His recording, made in four tracks on half-inch tape at 15 IPS, was called an experimental recording, dubbed “a practical test of tetrahedral ambiophony.” (Decades ago, JVC released an unrelated four-channel version on a quad vinyl disc, but the frequency response and dynamic range limitations inherent to that format resulted in the Toshiba multi-channel recording sounding noticeably worse than the standard Unicorn stereo recording.) 

HDTT is very pleased to be releasing the unheard Bruck recording, the preparation of which is an ongoing work in progress. This release will also include Strauss’ Death and Transfiguration, which was recorded at the same sessions. The Mahler and Strauss pieces will be released in a High Resolution format including all four channels, as well as in a two-channel stereo mixdown. The basic sound quality captured by Bruck’s outstanding skill as an audio engineer improves considerably over the duller, compressed sound heard in the Unicorn recording, and HDTT will of course maintain the very wide dynamic response present in the Bruck recording (even though that will require a warning to listeners!).

Bruck used a unique mic set-up that captured the sessions with remarkably well-focused clarity. No fan of multi-miking, he used a fixed array of four hyper-cardioid Schoeps mics placed in the form of a tetrahedron with a mic at each corner thereof, about one meter apart, with cancellation at about 150° off axis, so the null of each microphone was at the maximum point of sensitivity of all the other mics. This array was situated relatively near the orchestra, which was basically seated normally, with the two front mics recording the left and right channels, another one facing the rear wall of the hall in the same plane, and the fourth pointed straight up at the ceiling. Unlike some recording set-ups that place mics in the rear of the space to capture ambient signals, Bruck captured the ambient hall sound as reflected back to the nominal listener’s ears from the hall’s rear walls and ceiling, with that nominal listener seated roughly in the center of the mic array. 

Consistent with Bruck’s philosophy of minimal miking, the four-channel result is a study in phase coherence, and that is obviously an essential ingredient for HDTT to maintain as much as possible without blurring it by applying any kind of injudicious audio manipulation. Because very few audiophiles have a full range speaker mounted on the ceiling directly above the listening space (except this writer, who mounted one there to listen to this recording in its original state), the “up” channel can be viewed as an alternate ambient signal for more conventional multi-channel listening, as human beings do not have strong directional recognition of sounds coming from directly above the head (as this writer has demonstrated for himself). But the “up” channel makes sense, as in every live concert setting, we are in fact hearing a good portion of the sound being generated by the performers on the stage as it is reflected to us off the ceilings of both the stage and the hall—remove those ceiling reflections from the performing space and you have basically destroyed its acoustics. (By way of example, acoustic tile on the ceiling of a performing space often makes it unsuitable for “acoustic” music performed without electronic amplification.) 

The distances that the direct sound and its reflections traveled to reach the mic array (i.e., the position of the nominal listener) are close to the same for the rear and “up” channels, having a similar amount of temporal delay as compared to the front channels (which we can observe on a DAW’s computer screen). Consequently, the “up” channel can be seen as representing an alternate “viewpoint” for hall ambience when heard in a listening room in the same horizontal plane as the other three channels. The effect of hearing all four channels is not to swamp the music in reverb but rather to add another dimension of clarity and realism, an effect that is clearly perceived when the rear channels are suddenly muted. The locational cues are of course still provided by the front channels, as the direct sounds presented by those channels arrive to the listener first. A personal observation is that the rear channels show a different decay pattern for some sounds than for others, as compared to the front channels, consistent with the fact that the rear channels consist of entirely reflected sounds that have undergone a different kind of diffusion—in short, the rear channels are not merely delayed carbon copies of the front channels. 

The particular challenge of preparing this project for release is that the session tapes recorded by Bruck have never been edited down into finished recorded performances. As a result, all of the many session tapes have been digitized in a high-resolution format by HDTT’s Bob Witrak on HDTT’s outstanding equipment. And it will be recalled that Mahler’s Third Symphony is the longest work in the standard orchestral repertory—the Unicorn recording of it runs to almost 98 minutes. The first movement alone exceeds 38 minutes.

Fortunately, we have Mahler authority Deryck Cooke’s typed notes from the original editing sessions, in which he assisted, but only for the last 5 of the 6 movements. For the long first movement, we have had to refer entirely to the Unicorn recording to determine what takes were used and what edits were made, which has been arduous because there were a great many takes spread over several sessions, not always done in sequence,  and a large amount of smaller editing consisting of individual notes or phrases. And of course we are also referring to the Unicorn recording to check the editing of the other movements as well. When this process is completed, the end result will reflect exactly the same editing as that for the recorded performance that was approved by Maestro Horenstein, although the recording itself is an entirely different one that has never been previously released.

Since I have listened to and worked with the unedited session tapes as recorded by Bruck, I can state that the recorded performance released by Unicorn is a testament to the great skill of those who carried out the original tape editing duties, as well as to the incredible skill, memory and persistence of the masterful conductor in charge. For example, the beauty and polish of the brass playing heard in the released recording, including the iconic trombone solos in the first movement, suggest a superb orchestra performing at its best. The reality of the brass playing heard in all the sessions belies any such impression--few brass passages survived without splices from various takes, which I speculate relates to the notion that the quality of the augmented brass players retained for the sessions was not on the same level as that of the regular orchestral players (the brass scoring for this massive, infrequently performed symphony calls for 8 horns, 4 trumpets with 2 more optional, 4 trombones and tuba). We are very grateful that digital editing today is far easier than editing tape was with a razor blade, fifty years ago—that was the only means available for editing at that time. 

Working on this project has also given both Bob and me enormous respect for the brilliant work of Jerry Bruck as a recording engineer. Bruck, who is thankfully still with us today, in good health, is himself a respected Mahler authority, and he was clearly a visionary who was way ahead of his time in past decades. Listening to the sound quality he achieved in this alternate recording of Horenstein’s Mahler Third from a half-century ago, it is hard to believe that the recording is that old (by comparison, the outclassed Unicorn recording is much more typical of a respectable recording produced in that era). Both Bob and I have marveled at how Bruck achieved such remarkable results using the technology available at that time—for example, the capture of such an enormous dynamic range in that pre-digital era without a hint of distortion (obviously based on a clear understanding of what was actually possible with analog equipment that many engineers did not have), or the pinpoint clarity of the sound-stage resulting from his “just right” mic placement and his precise understanding of what his mics could achieve. Bruck believed in “getting it right” without compromise.

Kalman Rubinson Music & Equipment Reviewer, comments after he listened to the sample:
“I found it quite thrilling to hear this familiar recording as with new ears! It was just so much cleaner and more open. I noted that it had a much wider dynamic range than the original.”

 A downloadable “sneak preview” sample is provided, consisting of the first half of the first movement as heard in the front two channels. The level may be perceived to be low, which is a function of the music being soft and the dynamic range large, so turn it up, but with the awareness that loud passages will be loud (HDTT is not about to compress the dynamics of this recording!). 

Download the sample here 1.1gb 24bit/192khz


The sample is stunning – amazing “air” and presence in the sound!


Ok—spring is past. I’d hoped to listen before hearing Payere’s MSO concert. What’s the deal?

Paul Jacob

So when is “soon”? Ditto the Leinsdorf Walkure, which has already been mentioned in Fanfare?

Nathaniel Brown

Will it be available in cd sacd format or only as download! Many thanks for your reaction in advance!

Maarten Brandt

Have to say (as a recording engineer) I’m suprised by the adverse criticism of the original Unicorn release. Though multitracked (the first classical LP to be recorded this way I believe) I thought it was a major acheivement, stunning sound and performance with a large dynamic range. And the CD transfer didn’t dissapoint either. Still, like others here, I’ll be most interested to hear this new discovery.

David Lord

Leave a comment