Review of KAB EQS MK12 remastering phono preamp
Reviewed by Gary Galo for AudioxPress magazine October 2003
Reprinted with permission from the publisher.

K-A-B Souvenir EQS MK12 Disc Remastering Preamplifier. K-A-B Electro- Acoustics, PO Box 2922, Plainfield, NJ 07062. 908-754-1479, FAX 908-222-3442, info@kabusa.com, www.kabusa.com.Warranty: One year.

K-A-B are the initials of Kevin A Barrett, proprietor of K-A-B Electro-Acoustics, a small audio company based in New Jersey specializing in analog disc playback.
K-A-B carries a full line of products for playing disc records of practically every type, including turntables, custom stylii, signal-processing devices, preamplifiers, and a wide variety of accessories. Their website is a "must visit" for any- one who collects 78s and other types of historical recordings.
The EQS MKl2, K-A-B's top-of-the-line stereo phono preamplifier, has been designed to provide user-selectable playback equalization for records pre-dating the RI.AA stan- dard, including 78-rpm discs, early LPs, and program transcriptions

FEATURES
The EQS MKl2 has two phono inputs, selectable at the front panel. Many collectors will have separate turntables dedicated to 78-rpm records and LPs- the EQS MKl2 allows both to be connected. You can adjust cartridge load- ing at the front panel with trimmer ca- pacitors that are continuously variable between 20pF and 200pF. A 12-position rotary switch selects the resistive loading, adjustable between l00 and 100k. Twelve equalization curves are sup plied to accommodate the myriad record- ing curves used throughout the history of the disc record. K-A-B calls this a "Chronologic Equalizer" since the push buttons are arranged as a record equalization timeline, advancing through the history of equalization curves from left to right.
The transition points for the curves are shown in Table 1. Barrett has based his selection of the various turnover points on practical ex- perience, as well as data supplied in the Radiotron Designer's Handbook (CD- ROM printed versions are available from Old Colony; Chapter 17 is required reading for those interested in these matters). Stanley Lipshitz's landmark Audio Engineering Society paper was used to calculate the filter designs (also other required reading). A discussion of the nature of disc recording equalization is beyond the scope of this review. For an overview, I suggest my paper "Disc Recording Equalization Demystified," reprinted in The LP is Back!. \2

In Table 1, f3, f4, and f5 refer to the three transition points, as labeled by Lipshitz, The low-bass turnover is f3 the bass turnover is f4 and the treble transition frequency is f5 The AC curve is for acoustic records, and AE is for very early electrical discs (such as Victor electrics from 1925 that still bear the wing-style acoustic labels), Curves E3, E5, and E7 are for the bulk of electrically recorded 78-rpm discs made from the mid-1920s through the late 1940s. For these three curves, K-A-B uses their unique "Fine Slope" high-frequency attenuation, rolling off at 3dB-per octave above a corner frequency of 2120Hz, and shelving at -10dB. Some collectors and transfer engineers prefer to think in terms of the attenuation at 10kHz, rather than the actual treble transition frequency.
Table 2 is a conversion chart which should prove helpful. A special CO curve complements the late Columbia 78-rpm record. The NAB curve (for National Association of Broadcasters) is the standard for 16" lacquer transcription discs (often incor- rectly called "acetates") used in the broadcast industry. Four 33 1/3-rpm curves cover the original Columbia LP, AES (Audio Engineering Society), the early Decca/London FFRR ("Full Frequency Range Recording"), and RIAA (the Recording Industry Association of America). \3 The twelfth and last curve on the EQS MK12 is flat, which means that it's not really a curve at all.
The flat position is useful for a couple of reasons. Acoustical recordings approximate a constant- velocity characteristic throughout their limited frequency range. A constant-ve- locity recording will yield a flat frequen- CY response when played with a magnet- ic cartridge (for further details, see my previously mentioned paper). Therefore, those inclined toward a strictly scientific approach to playback equalization often prefer a flat (i.e., non- equalized) playback. Many collectors, myself included, prefer adding a bit of upper-bass/lower-midrange warmth to acoustic records. If the flat position is used, the warmth can be added with an equalizer, preferably a parametric

EQUALIZATION
Transfer engineers who use digital processing (such as CEDAR) for removing noise on 78-rpm recordings sometimes find that a flat initial playback allows the digital noise removal to work better, before the recording curve is applied. The flat position can be extremely useful here. The disc is played without equalization, then fed to the digital processor. The EQS MK12 also has a line input, just ahead of the equalization circuitry, which allows you to apply the recording curve at line level, after the digital pro- cessing has been done. Some engineers will tell you that you can apply the play- back curve in the digital domain, and the equalization capabilities of many computer-based digital processors and editors will allow you to do this.
There's a flaw in this approach, however. Among the virtues of digital equalization is the lack of nasty phase shifts inherent in analog filters. In the case of playback curves, the lack of frequency-dependent phase shifts presents a problem. Most readers know that the playback curves provide proper frequency equalization for the record. What is generally ignored-probably because it happens automatically-is the fact that the play- back curves also provide correct phase equalization. The filters used to pro- duce the recording curves in the first place caused phase shifts. When complementary analog play- back equalization takes place, the phase response of the record is also corrected. Ideally, a properly equalized record will have flat frequency and phase response (the fact that we can geta credible square wave off an LP test record is proof that we can come pretty close when everything is working properly). To achieve this, old-fashioned analog playback equalization is the only way to go; digital filtering won't accomplish this task.
It should certainly be possible to design digital filters to mimic both the phase and frequency characteristics of analog filters, but no one has yet done this commercially for playback equalization of disc records, as far as I know.
The line input on the EQS MK12 is applied directly to the passive equalization circuits, bypassing the first gain block. The equalization circuits apply a low impedance load-around 2000Q-to the source, which should not be a prob- lem for most professional gear with very low-Z outputs. If it is a problem, K-A-B offers a version of the preamp with a high-impedance line input board this option adds $100.

After the Chronologic Equalizer comes the processor section, which has a fourth-order (24dB/octave) rumble filter with a -3dB point of 30Hz, plus a processor loop for connection of an equalizer or other signal-processing device. The stereo/mono switch is used in conjunction with the lateral/vertical switch and the mono mix controls. Most monaural recordings are laterally cut. Vertically-cut recordings include all cylinders, Edison Diamond Discs, most Pathe discs, and some others.
When a stereo cartridge is used to playa laterally-cut monaural record, the left and right channels are summed to mono "in phase." The vertical component of a monaural record- ing, which is almost entirely noise, is cancelled. When the same cartridge is used to playing vertically-cut recording, the polarity of one channel must be reversed before the two channels are summed to mono. In this case, it is the lateral component (again, mostly noise) that is cancelled.
The mono mix control is like a balance control, and determines the amounts of left and right information that are summed. Under ideal circumstances, the left and right signals will be identical in level and phase. Old records are far from perfect, and many do not have identical characteristics in the two groove walls, even though they were recorded with a monaural cutter. The best way to adjust the mono mix control is to put the lateral/vertical switch in the position just the opposite of the way the record was cut. Put in the vertical position for a lateral recording, and adjust the mono mix control for a null in the signal. A "perfect" record will yield complete cancellation, and only noise will remain. Then put the switch back in the lateral position for playback.
Most laterally-cut records have some vertical component, and vice-versa, so complete cancellation won't always occur-you simply adjust for the lowest signal level. Often old records have different levels of wear on the two groove walls. The mono mix control allows you to select either groove wall, or any mix of the two. The EQS MK12 also has a 12-position gain control, allowing the preamp to be interfaced with a variety of consumer and professional equipment. K-A-B points out that this switch should not be used as a volume control.
The EQS MK12 is intended to be used with a pre- amplifier, integrated amplifier, or other control center. Both unbalanced RCA and electronically-balanced 1/4" phone (tip-ring-sleeve) outputs are provided. The RCA jacks are high-end gold-plated, Teflon™ insulated types.

DESIGN PARTICULARS
Although schematics were not supplied with my review sample, a look inside revealed the general nature of the circuit topology (Photo 3). Each channel of the phono preamplifier uses switchable, passive equalization networks, situated between two gain modules, which are discrete and consist of five transistors each. One percent-metal-film resistors, plus Wima and Panasonic polypropylene capacitors, are used in the filter networks, and throughout the preamp
The stereo fourth-order rumble filter is built around a National Semiconductor LF347BN quad J-FET op-amp. The line stage uses a pair of National LM837N low-noise, high-output current, bipolar quad op amps one of these is used for polarity inversion for the lateral/vertical switching, and the other is configured as a differential output buffer amplifier. The output gain switch is not an attenuator; the ten-position switch is configured to change the gain of one of the line stage op-amp sections, by switching parallel feedback resistors. K-A-B also switches parallel capacitance, in order to maintain the same bandwidth at the different gain settings.
In a rather radical departure from today's norm in solid-state audio design, K-A-B has employed a single- ended power-supply topology. The supplied "wall-wart"-type power transformer is rated at 24V DC at 400mA, but even under the load of the preamp the transformer output actually measures 28V DC. Inside the preamp, this raw DC supply is fed to an L/C choke-input filter, and the filter output feeds an LM317 pre-regulator with the output voltage set at +22V.
The line stage and rumble filter op amps have their own regulators, a pair of 7818 three-terminal types. Each of the discrete gain modules in the phono preamplifier-four total for the two channels-has its own on-board 7818 regulator. K-A-B uses the term "Polar Stable" to describe their single-ended approach. In an answer to an e-mail inquiry I posed, Barrett noted that
"The polar design dictates that along the signal path there will always be a bias voltage present. I have come to regard this bias as having a stabilizing influence on the circuitry and components. I believe it plays an important role in the sound of my designs. Tube designs tend to share this similarity also."
In a single-polarity power-supply topology the signal path will, ideally, be at a potential equal to exactly half of the supply rail. For a variety of reasons, this is not always the case. If the signal path deviates from this ideal, asymmetrical clipping will occur causing a reduction in headroom (one side of the waveform will clip prematurely). K-A-B has solved this problem by adding a fixed dc bias, supplied by a 78LO9 IC regulator. The bias holds the signal path-specifically the outputs of the various amplification stages-exactly mid-way between ground and the supply rail at all times, ensuring symmetrical clipping and maximum headroom.
The EQS MK12 is capacitor-coupled, typically with Panasonic HFS electrolytics bypassed with polypropylene film capacitors. The main output coupling capacitors appear to have a triple-film bypass. The switching in a preamp with a single-ended power supply can produce clicks and pops if care is not taken to keep the outputs of coupling capacitors at OV potential with resistive loading. All switches in the EQS MK12 are completely silent in their operation, including the much used equalization selectors. The construction quality is excellent.

MEASUREMENTS
I made all measurements on my Sound Technology 1700B analyzer.
Figure 1 shows the frequency response in the flat position. The EQS MK12 circuitry has been designed for wide bandwidth, with the response at -2dB at 10Hz and -3dB at 100kHz.
My measurement showed the -3dB point for the rumble filter to be at 28Hz, rolling off at 24dB/octave below that frequency. I used the Jung/Lip- shitz Inverse RIAA Network (TAA 1/80) to measure the RIAA accuracy, with 1Voutput at 1 kHz as my OdB reference. K-A-B does not specify the RIAA response- my measured results are shown in Fig. 2. K-A-B has built some infrasonic rolloff into the RIAA circuit, so the low end is at -0.7dB at 20Hz. This is a more conservative infrasonic rolloff than the IEC amendment to the RIAA specification. I have never liked the IEC call for a -3dB point of 20Hz (corresponding to a time constant of 7950us). I have tried this in my own preamp and found that the degradation of the bass was quite audible. K-A-B's solution is more sensible, putting the -3dB point nearly an octave lower than IEC.
The RIAA performance is ideal throughout the critical midrange, measuring ± 0.1dB from 500Hz to 20kHz, and ± 0.25dB from 40Hz to 20kHz. There are no inverse networks for the ten remaining curves included in the EQS MK12, so I simply measured the actual response of each one. I set my OdB reference at 1kHz, with the pre- amp output driven to 0.5V out (corre- sponding to a phono input level of 5.3mV; the 1kHz gain of the preamp measures 39.5dB). I carefully moniitored the preamp's output on an oscilloscope to make certain that the preamp was not clipping in the low frequencies. The six 78-rpm curves are shown in Fig. 3, and the four 33 1/3-rpm curves are shown in Fig. 4. All measurements showed that the EQS MK12 performs as specified. When examining Figs. 3 and 4, bear in mind that I used 1kHz as a convenient reference for all ten curves. In Fig. 3, the E3, E5, and E7 curves show the K-A-B Fine Slope technique in the treble region, shelving at 10kHz. K-A-B specifies f3 for the LP curve as 50Hz, which made me initially suspicious. The original Columbia LP curve put the low-bass turnover at 1OOHz. But, the low-frequency gain stop incorporated by K-A-B for the LP curve, combined with f3 of 50Hz, produces the proper result, as Fig. 4 shows.
All frequency response measurements were essentially identical in both channels any differences were beyond the resolving capability of my test equipment. I measured total harmonic distortion in the flat position, with an input level of 11mV, which produced an output level of 1V. Figure 5 shows THD to be around 0.03% across most of the spectrum, rising to 0.05% at 20kHz. Noise dominated the distortion products at lower frequencies, with some second harmonic introduced at 10kHz and 20kHz. There were no higher order distortion components. Changing to RIAA equalization lowered the THD to 0.01% at 1kHz, 10kHz, and 20kHz, with the distortion products consisting entirely of noise. Two-tone SMPTE IM distortion measured 0.011%. All distortion measurements were the same in both channels. In the flat position, the output before clipping was 5.1V unbalanced, and 10.2V balanced. In the flat and RIAA positions, the maximum input level just before clipping was 62mV. Noise measured 71dB below 1V in the flat position with the input shorted; the high-frequency rolloff of the RlAA curve reduced the noise to 78dB below 1V.

THE SOUND
To evaluate the basic sonic quality of the EQS MK12, I used several of my ref- erence stereo LPs (Table 3), all cut with RlAA equalization. My reference LP playback system is my own belt-driven, AR/Menill-based turntable fitted with a Grado Signature tonearm and an Adcom XC-MRII high-output moving coil cartridge. The turntable is powered by an electronic speed control of my own design (Photo 4).4
The EQS MK12 proved to be a fine- sounding performer offering a warm and detailed sonic presentation. The treble region is silky and smooth, lending itself to fatigue-free long-term listening. Overall, I found the warmth and liquidity in the sound to be somewhat tube-like, though not over-ripe. The pre- amp offers a hint of the euphonic qualities of a good tube preamp, but it is not overly colored. Soundstaging is a bit narrower and shallower than my reference preamplifier (my extensively modified Adcom GFP-565), but localization is generally very good. I had no difficulty following the subtle stage movements of the singers in the Culshaw-produced Wagner and Strauss recordings conducted by Georg Solti.
The bass region is a bit reticent, lacking the weight and impact of my reference preamp, but it is clean and well defined. Overall, the EQS MK12 is a solid performer with stereophonic/high-fidelity material, offering satisfying musical performance.
Over the course of several months of listening, I played literally dozens of pre-RIAA recordings, mostly 78-rpm discs, but also including early 33 1/3- rpm material. My 78-rpm playback system includes a Technics SP-15 turntable and an SME 3012R tonearm mounted on a custom base with isolation feet, plus a Stanton 500A-series cartridge with a variety of truncated stylii (Photo 4).
I also use this turntable for some 33 1/3-rpm recordings, including 16" transcriptions, Vitaphone sound- tracks, and early-1930s Victor LPs. Table 3 contains favorite recordings that I found especially useful, but it is only a partial list. The EQS MK12 did an excellent job with the discs I auditioned.
Overall, the EQS MK12 easily outperformed my modified McIntosh 0-8, which has been my reference 78-rpm preamp for over 15 years.5 The EQS MK12 offers a cleaner and more detailed sound than the 0-8, with a more open and transparent treble region. I found the EQS MK12's curves to be intelligently chosen, with the AC curve offering just the right amount of Warmth for most well-recorded acoustic discs. The moderate high-frequency rolloff reduces surface noise without cutting into the limited treble region on acoustical recordings. It is rare to find an electrically recorded disc that won't offer musically satisfying performance with one of the supplied curves. If you've never heard 78s played with proper equalization, the EQS MK12 is likely to be a revelation. The lateral/vertical switching and mono mix controls worked superbly. A proper mix can make a great difference in the sound of 78s, and I consider this control essential for optimum playback. There were times when I wished that the rumble filter began at a higher frequency. Many 78s have audible rum- blings in the 40-60Hz region, including the Meistersinger Quintet recording on my list. A higher corner frequency would admittedly compromise many recordings, so I really can't fault the designer's choice. Serious collectors will probably find a parametric equalizer an essential accessory.
It is often desirable to put the processor loop after the lateral/vertical switch- ing and mono mix. If a stereo equalizer is used ahead of these controls, you can't get a proper null with the mono mix controls unless the two channels of the equalizer are precisely matched. A better solution is to put the equalizer at the output of the EQS MK12. In this case, you need only a mono equalizer, which can then feed your control preamp.
I suggest the Behringer Ultra- Q@ Pro PEQ2200 as a cost-effective solution, or the Symetrix 551E for greater flexibility K-A-B sells the Symetrix, and most pro-audio dealers carry both). I know a number of professionals involved in commercial transfer of historical recordings, and many insist on separate control of the bass turnover and treble rolloff frequencies, which the K-A-B doesn't offer. There are many electrically-recorded 78s with 6dB/octave high-frequency preemphasis curves that are not exactly complemented by the "Fine Slope" rolloff in the EQS MK12, though I found that the "Fine Slope" curve worked well with most of the 78s I played. With the addition of a parametric equalizer, however, you can make sufficient adjustment to accommodate a variety of 78-rpm records. The AES curve, with its 6dE/octave slope, is also useful on many 78-rpm discs.
CONCLUSION
The K-A-B EQS MK12 is an excellent solution to the problems facing collectors of historical recordings. A high-quality preamplifier with a variety of equalization curves will breathe new life into old recordings, and the EQS MK12 is proba- bly the best all-around product of its kind currently available.\6 Complemented with a good turntable, cartridge, and stylii, the EQS MK12 makes a fine playback system for 78-rpm and other vintage material. The AC and FLAT settings should also work well with cylinders.7 The EQS MK12 comes with a helpful instruction booklet and a chart offering recommended equalizer settings for a wide variety of recordings.
If you are a serious collector, the EQS MR12 deserves serious consideration.

TABLE 1
MANUFACTURER'S SPECIFICATIONS
Input Section: Input capacitance. 20-200pF continuously variable
Input resistance. 100-100k in 12 steps
Input sensitivily. 11mV @ 1kHz, RIM, for 1V output
Input overload. 66mV @ 1 kHz, RIM, for 5.9V output
Fixed front-end gain: 36dB @ 1 kHz, RIAA, gain set to "0"'
Chronologic Equalizer:
Curve     f3     f4     fs      LF Gain Stop 2/
AC        50     500    5000     Yes (+10dB)
AE        20     200    2120     No
E3        30     300    2120 1/  No
E5        50     500    2120 1/  No
El        70     700    2120 1/  No
CO        30     300    1590     No
NAB       40     400    1590     Yes (+17dB)
LP        50     500    1590     Yes (+ 13.5dB)
AES       40     400    2500     No
FFRR      30     300    2120     Yes (+17.5dB)
RIAA      50     500    2122     No
FLAT      0      0      0        No (Gain fixed)
1/ K-A-B fine slope high-frequency rolloff, 3dB/octave
   shelving at-10dB.
2/ Low-frequency gain stops limit the total bass boost to
   the figure stated; ref. OdB @ 1kHz.
Rumble Filter: Corner frequency. 30Hz Attenuation 24dB/octave
Output Stage: Active balanced tip-ring-sleeve: 12V RMS maximum
RCA single-ended: 6V RMS maximum
Distortion and Noise (Ref. 1V out): THD: <0.05% IMD: <0.05% S/N. >-75dB
Physical Specifications: Dimensions (W x H x D): 19"x 1.75" x 8.25"
Weight: 5 Ibs Shipping weight: 8 Ibs
TABLE 2
TREBLE ROLL-OFF CHART
10kHz Attenuation Transition Frequency (and Time Constant)
-5dB 6800Hz (23.41us)
-8.5dB 4056Hz (39.24uS)
-10dB 3333Hz {47.75uS)
-10.5 3128Hz {50.88uS)
-12dB (AES) 2595Hz (61.33uS)
-13.73dB(RIAA) 2122Hz(75uS)
-14dB 2036Hz(78.17us)
-15dB 1807Hz (88.08uS)
-16dB(NABand 1591.55Hz{100.0uS) Columbia LP)
-20dB 1005Hz {158.36us)
The formula for converting dB at 10kHz to the -3dB frequency was generously provided by aX regular contributor G.R. Koonce.
REFERENCES

1. Lipshitz, Stanley, "On RIAA Equalization Networks" Jounal of the Audio Engineering Society, June 1979

2. Gale, Gary, Disc Recording Equalization Demystitied," The LP is Back!, Peterborough, NH, Audio Amateur Press. 1999 (available from Old Colony). The artiere is based on a paper given at the Association for Recorded Sound Collections conference in 1996, originally published in the Fall 1996 issue of the ARSC Journal.

3. RIAA has a website, www.riaa.org, but you won't find the RIAA LP curve here, just a link to National Semiconductors dataeheet and audio application note for the LM833 opamp.

4 Galo, Gary, AR System Drives New Turntable,". Audio Amateur 3/85; An Electronic Speed Control," Audio. Amateur 1/86; and The Belt-Dnven Turntable ' Revisited," Audio Amateur 3/88. These articles have been reprinted in The LP is Back!.

5. Galo, Gary, 'A Preamp for Vintage 78s,".Audio Amateur 1/85. Reprinted in The LP is Back!. My C-8 has been modified beyond what was described in this article.

6. The FM Acoustics Resolution 222 features separate control of bass- and treble-turnover frequencies, but the lack of a flat position and mono mix control makes it less useful for 78-rpm discs, and its $18,500 price tag puts it out of the reach of all but the affluent.

7. There's very little equipment available for electronic playback of cylinders. A couple of interesting products can be found on the website of Nauck's Vintage Records at www.78rpm.com. Their Advanced Cylinder Technology Reproducer houses a Stanton 500 carfridge and can be fitted to a number of original. Edison. cylinder phonographs.
TABLE 3
RECORDINGS USED FOR THIS REVIEW
RIAA STEREO LPS:

Wagner: Die Walkure--Ride of the Valkyries. Los Angeles Philharmonic Orchestra conducted by Erich Leinsdorf. Sheffield Labs Direct-to-Disc LAB-7 (Pressed in the US).

Rimsky-Korsakov: Scheherazade, Op. 35. Chicago Symphony Orchestra conducted by Fritz Reiner. Chesky RCA.

Wagner: Der Ring des Nibelungen, especially "Siegfried's Death and Funeral March" from Gotterdammerung (Side 11), and the "Forging Scene" from Siegfried (Side 3). Birgit Nilsson, Wolfgang Windgassen, et al. Vienna Philharmonic Orchestra conducted by Georg Solti. Decca 6.35500 (Mastered and pressed in Germany by Teldec).

Strauss: Salome, especially "Wird dir nicht bange, Tochter der Herodias" through Jokanaan's descent into the cistern (Side 2). Birgit Nilsson, Eberhard Wachter. et al. Vienna Philharmonic Orchestra conducted by Georg Solti. London OSA-1218 (Mastered and pressed in England by Decca).

Stravinsky: Le Sacre du Printimps. Cleveland Orchestra conducted by Lorin Maazel. Telarc DG 10054.

PRE-RIAA HISTORICAL RECORDINGS:

Schoenberg: Gurrelieder-Lied der Waldtaube. Martha Lipton, mezzo-soprano. Philharmonic-Symphony Orchestra of New York conducted by Leopold Stokowski. Columbia ML-2140 (10" LP; Rec. 1949).

Leoncavallo: Oagliacci - "Vesti la giubba." Giovanni Martinelli, tenor. Vitaphone Orchestra conducted by Hermann Heller. Vitaphone Soundtrack Disc, Matrix No. 300107 (331/3-rpm: vinyl pressing from the original metal part; Rec. 1926).

Cohan: Over There. Enrico Caruso, tenor. Victor 87294 (Acoustic; 78-rpm; "Wing" label; Rec. 1917).

Wagner: Lohengrin Mein lieber schwan!" Jacques Urlus, tenor. Edison Diamond Disc 83017-R (Acoustic; 80-rpm: vertically-cut: Rec. 1915).

Weber: Oberon-Overture. Berlin Philharmonic Orchestra conducted by Arthur Nikisch. HMV 1040 (78-rpm; acoustic; vinyl pressing from original G&T metal part issued by Symposium).
Beethoven: Symphony No. 7 in A Major. Philadelphia Symphony Orchestra conducted by Leopold Stokowski. Victor Set M-17 (78-rpm; 'Scroll label: Rec. 1927).

Strauss: Also Sprach Zarathustra. Boston Symphony Orchestra conducted by Serge Koussevitzky. Victor Set M-257 (78-rpm; 'Scroll Label; Rec. 1935).

Verdi: Otello-'Dlo! ml potevi scagliar' (sung in German). Lauritz Melchior, tenor. New Symphony Orchestra conducted by John Barbirolli. HMV D2037 (78-rpm: British pressing: Rec. 1930).

Wagner: Die Meistersinger von Numberg-Act III Quintet. Elizabeth Schumann, Lauritz Melchior, Friedrich Schorr, Ben Williams, and Gladys Parr. London Symphony Orchestra conducted by John Barbirolli. Victor 7682 (78-rpm; "Scroll" label; Rec. 1931).

Beethoven: Fidelio-Gott, welch' dunkel hier.' Helge Roswaenge, tenor. Berlin State Opera Orchestra conducted by Bruno Seidler-Winkler. HMV D.B.4522 (78-rpm: British pressing; Rec. 1938).

Chopin: Ballade No. 4 in F-minor. Alfred Cortot, pianist. HMV D.B.7589-7590 (78-rpm; British pressing; Rec. 1933). Wagner: LohengrinPrelude to Act I. Philharmonic-Symphony Orchestra of New York conducted by Arturo Toscanini. Victor Set M-308 (78-rpm; Rec. 1936).