Although these two recordings were never meant as companions (unlike the Brahms violin concerto recorded by Oistrakh with Szell three days after the Double Concerto, respectively 12-13 and 16 May, 1969, Brahms: Violin Concerto / Violin Sonata No. 3), the pairing is logical. The two compositions are usual and logical bedfellows on disc because of their unusual array of soloists, the soloists here are shared (with the addition of Richter in Beethoven's triple concerto) and the Beethoven was recorded only a few months after the Brahms, on 15-17 September 1969. They are also excellent versions, although Oistrakh's earlier recordings of both compositions for EMI (Beethoven in 1959 with his usual trio partners of those years Lev Oborin and Stanislav Knushevitzky under Malcolm Sargent, Brahms in 1956 with Pierre Fournier under Alceo Galliera, both with the Philharmonia), while not on a par sonically, were even better interpretively: Triple Violin Concerto.
The highly controversial English music critic Norman Lebrecht has given pride of place (so to speak) to Karajan's Triple Concerto in his list of "The 20 Worst Recordings Ever Made" - and it is pure crap, based on gossiping rather than listening, and apparently motivated by his general hatred of Karajan rather than by any fair hearing of the recording (see my review of Lebrecht's book for more, The Life and Death of Classical Music: Featuring the 100 Best and 20 Worst Recordings Ever Made). Karajan and his Soviet dream team expound an expansive and lyrical view (which doesn't preclude vigor in the tutti and playfulness in the interplay between the soloists), unfolding very naturally, almost nonchalantly at times in the first movement, and with a fine mood of playfulness in the finale. I'm usually a staunch advocate of flowing tempos in Beethoven's "slow" movements (alhtough, here, it is indeed a "Largo" and not the customary "Andante"), but in this instance I must lay down all arms in front of Karajan: here, his time-suspended tempo(and the soloist's ability to sustain it) is ineffably beautiful. I could fault Karajan and the sonics on a few details (wayward, unstable tempo in the first bars; a certain lack of crispness in the strings: Karajan doesn't always seem to think that a staccato mark should be played staccato; the extremes - basses and sometimes horns and oboes - are covered in the tutti, which is inherent to the fact that the body of strings is basically too large for Beethoven's scoring; the string tutti sound too thick and distant in relation to the soloists in the finale, they sound as if Karajan thought he was recording a Symphony; in that respect, recent versions, like Harnoncourt's, Beethoven: Triple Concerto; Rondo in B flat; Choral Fantasy, achieve a much better balance; but then, there aren't so many tuttis in the finale), and those details would make me give preference to the 1959 recording (Oborin and Knushevitzky have little to envy from Richter and Rostropovich in terms of technical authority and interpretive imagination, and Oistrakh's tone is even more beautiful than in 1969) if its sonics had been comparable. But it has significant tape hiss and the soloists aren't as clearly and brilliantly recorded as in 1969. Anyway these small blemishes do not diminish the value of this later version, that lies on its stellar soloists even more than on its conductor (and credits to the conductor anyway for the sublime Largo): not that other approaches, more urgent, can't be conceived, but it is truly, in its own style, a classic.
It is fascinating to follow the artistic development of Oistrakh through his successive recordings of Brahms' double concerto. It confirms the general rule (but not an absolute one: Toscanini and Heifetz are notable exceptions) that performers mellow and pace down as they age. I haven't heard Oistrakh's 1946 (some sources say 1947) recording with Milos Sadlo and the Prague Radio Orchestra under Karel Ancerl, but I have his second studio recording, with Knushevitzky and the Leningrad Radio Orchestra under Karl Eliasberg, which discographic sources date from 1948 (although on my French Chant du Monde CD, volume 12 of their David Oistrakh Edition released in the late 1980s and listed nowhere, it sounds much better than what you'd expect of a Soviet recording from that vintage - see my review of David Oistrakh Plays Doubles for more details on that dating issue). While already displaying Oistrakh's unique lyricism, it is also faster than his subsequent studio recordings, and (thanks also to Eliasberg whipping his orchestra to great intensity), it is also more fiery, almost Heifetzian even, with an undertone of unrest and inquietude in the central andante. The 1956 recording with Galliera has more of a classical poise, Galliera goes slightly more for power and grandeur rather than bite and drive (which doesn't mean that it is dispassionate in any way), and the andante, taken at a very held-back tempo, is imbued with a uniquely tender lyricism.
Szell is even more expansive in the outer movements (compare his first movement 16:43 to Galliera's 16:04 and Eliasberg's 15:10) , but he avoids any impression of sluggishness thanks to the great mass and power of his orchestral outbursts, and the bite of his accents and staccato. Still, despite the 1956 early stereo with limited spread and significant tape hiss (especially when you crank up the volume), I would give the edge to the earlier version.
The 1969 recording isn't ideal either; it was the first recording made by the EMI engineers of the Cleveland Orchestra, and the orchestra doesn't have the clarity and transparency you'd expect of a 1969 recording, with even a hard edge in the tutti. There is also a thickness and lack of bounce of those tutti in the finale (try at 1:00), but that must be ascribed to Szell rather than to the sound engineer, and in that finale one may prefer Galliera's marginally stronger drive and bite. As recorded in 1969, Oistrakh doesn't quite reproduce either the angelic beauty and sweetness of tone of his former 1956 self. For all Oistrakh's and Rostropovich's involvement, there are some spots in the first movement where one also feels them tempted to over-express. But it is especially in the andante that Szell and his soloists fail to entirely recapture the special magic of the 1956 version.
Needless to say, all these remarks are only comparative. This is an outstanding version in its own right, and a safe recommendation for those who might be bothered by the more rustic stereo and pronounced tape hiss of the 1956 recording. But those who have never heard the 1956 Brahms andante must know that they will have missed something.