If you’ve spent much time looking at the etched label areas of American Berliner gramophone discs, you may have noticed the cryptic alphanumeric markings that often turn up near their spindle holes, most commonly at three o’clock (but sometimes at nine o’clock). Paul Charosh has this to say about them in Berliner Gramophone Records, American Issues 1892-1900 (Westport: Greenwood Press, 1995), xxii:
Fractions and mathematical expressions. There are none on these discs, but markings to the right of the center hole resemble them…. The compiler believes they are manufacturing or related codes of some sort, useful to the recording technician. Their precise significance is unknown and they have been ignored in this volume.
I’ve been curious about these mystery markings for some time now. After all, if they really were “useful to the recording technician,” as Charosh surmises, I supposed that deciphering them might offer valuable insights into early recording industry practices. And I believe I’ve now cracked at least part of the code, thanks to a critical mass of data provided by high-quality photographs of discs in the David Giovannoni Collection, which is the source of all the illustrations of Berliner markings shown below.
In the interest of keeping things manageable, I’m going to restrict this particular blog post to a limited subset of Berliner gramophone discs, albeit a fairly large subset. In the spring of 1899, Berliner inaugurated a new system of sequentially assigned catalog numbers with a leading zero—01, 02, 03, etc.—which supplanted an older block numbering system and remained in use until the end of his operations in the United States in May 1900. (Specific takes from the 0-prefix period are conventionally designated ts, with a take symbol at six or nine o’clock, or nt, with no take symbol.) Because the technical inscriptions from this period strike me as comparatively straightforward and consistent, I’ll start by analyzing them here with the idea of working backwards to make sense of earlier markings in a follow-up post.
Technical inscriptions on discs with the lowest 0-prefix catalog numbers most often take the form letter-number-letter, with the first two parts written by hand and the third part stamped.
The frequent presence of hyphens implies that the letter-number-letter combination was meant to work together as a composite technical identifier. However, the final letter is often stamped much further to the right of the handwritten part of the inscription than in the examples shown, and for catalog numbers starting in the 0130s (and post-April 1899 remakes of lower numbers) we find it relocated to the bottom of the label area, underneath the stamped catalog number. The meaning of the stamped letters is already well-known to Berliner researchers, apparently based on inference rather than explicit documentation: they’re genre markers, with J signifying “baritone” and U signifying “talking” in the examples illustrated above. Moving the genre marker away from technical inscriptions of this sort to the bottom of the label area left a pattern of one letter plus one number, and these turn up written together in a variety of styles.
Along with combinations of one letter and one number, other more elaborate cases can be found with additional numbers thrown into the mix. Sometimes there are two numbers, written either in sequence or one above the other.
Starting in January 1900, we occasionally also find three numbers, always written with two of the numbers grouped together on the opposite side of a line from the other number, resembling a more complicated fraction.
So what do these inscriptions mean? If they do indeed represent technical metadata, we can identify two major variables which recordists of the period would likely have wanted to track: the choice of recording horn and the choice of sound-box and diaphragm. Both of these parameters varied with the type of content being recorded, so the temporary connection of such metadata with stamped genre markers would have made sense; taken together, these codes would have indicated what type of content had been recorded as well as how it had been recorded, enabling recordists to assess the results of applying particular technical strategies to particular subjects.
Only a single sound-box or diaphragm could have been used for recording at any given moment, which suggests that this variable must have been tied to the element in the inscription that doesn’t ordinarily seem to allow for repetition: the initial letter. For technical inscriptions from the 0-prefix period, then, my hypothesis is that initial letters refer to sound-boxes with specific diaphragms mounted in them. The letters attested so far for the 0-prefix period are A, C, D, E, G, H, J, K, P, R, S, T, Y, and Z. Most of these turn up multiple times, but I’ve only spotted one instance of T (0151 ts), so other “missing” letters might simply be very uncommon. Different sets of letters are also common in different periods; thus, A, C, G, and H are the most common letters in early to mid-1899, whereas I’ve seen nothing but P, R, and S on recordings made during 1900. This pattern would be consistent with advances in the art through which newer sound-boxes superseded earlier ones. I don’t see any obvious pattern connecting position in the alphabet to characteristics of sound-boxes, so maybe the letters were simply assigned in the order the sound-boxes were built.
I hypothesize further that numbers refer to recording horns, with combinations of multiple numbers documenting recording sessions at which multiple horns were linked together with tubing and used simultaneously. The numbers attested for the 0-prefix period so far are 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 12½, 13, 14, 15, 17, 18, 19, 20, 31, and 32. The 12½ suggests that these numbers must refer to some kind of meaningful measurement or range of values rather than to an arbitrary sequential numbering system, as does the conspicuous gap between 20 and 31. One possibility is that the numbers refer to the lengths of horns in inches. The number 2 turns up by itself mostly on speaking records by Cal Stewart and George Graham, and in one case on a soprano solo by Florence Hayward; I suppose these discs might have been taken with extremely short two-inch horns, with the performer’s mouth stuck right up into them. Large ensembles tend to be associated with higher numbers, which would correspond to longer horns. That said, I don’t know why Berliner would have had a (nearly?) complete set of horns in one-inch increments all the way up to 20 but nothing between 20 and 31. But this is the most plausible interpretation I can come up with at the moment.
Targeted listening comparisons might be able to confirm or disconfirm these interpretations. For instance, we could compare George Graham’s “Police Court” (0643 nt, R4) with his “Talk on Drinking” (0644, R32), which I’m predicting would have been recorded using horns of lengths at nearly opposite ends of the spectrum; and we could compare Arthur Collins’s “I Couldn’t Stand To See My Baby Lose” (0921, R11) with his “You Told Me You Had Money In the Bank” (0922, S11), which I’m predicting should have been recorded using different diaphragms with different degrees of sensitivity.
Up through the end of 1899, two-number technical inscriptions are always associated with ensembles: orchestras, bands, vocal and brass quartets, and vocal and instrumental duets—all cases in which it would have been advantageous to aim two horns towards different parts of the group and to route the sounds to the sound-box through a branched tube or some other means of “mixing” them. If the elements in the ensemble are roughly interchangeable, the two numbers tend to be the same, as with banjo duets marked P 9/9 (0816 nt, 0818 2); I’d interpret that as a combination of two identical “9” horns, one aimed at each banjo. For ensembles that are more diverse, the numbers may be the same—e.g., two “17s” for a band record—or different. Vocal quartet records sometimes combine two identical numbers (e.g., K 8-8 on 0382 nt), but often they instead combine a “2” and a “10,” which might reflect the use of a short “2” horn to foreground the lead tenor and a longer “10” horn to pick up the rest of the quartet (e.g., 0286 nt). Or maybe the goal in those cases was to capture a piano accompaniment; again, targeted listening comparisons might be able to clear this up. In early 1900, the scope of two-number technical inscriptions expands to take in a few soloists (banjo, violin, vocal) accompanied on piano, always combining a “2” with a higher number (10, 11, 31). Here too, the combination of a very short horn for a soloist with a longer horn for a piano accompanist could have been a handy strategy for mixing the signals at desired levels.
The three-number technical inscriptions of 1900 are associated either with large bands and orchestras or with unusually complex subjects such as minstrel records, dances with figures called, and vocals with orchestra accompaniment. If my interpretation is correct, these can be understood as referring to sessions in which three horns were used, with two successive branchings of the signal chain. One telling example of this type is 01061 take 2, “The Blue and the Gray,” sung by baritone Arthur Collins with orchestra accompaniment, technical inscription R 11 / 17-17. R 11 is a common technical inscription for baritones with piano accompaniment (e.g., Collins himself on 0753 nt). On the other hand, the combination 17-17 is commonly associated with band records (e.g., R 17-17 on 0503 nt). So if my interpretation is correct, R 11 / 17-17 would represent a case in which one of Arthur Collins’s usual horns (“11”) was linked to a combination of two horns ordinarily used together for taking large ensembles (“17-17”) for an occasion when he was singing with an orchestra rather than with the usual piano accompaniment.
And here’s a comparable diagram of the arrangement I’d infer for 0912 nt, a recording of dance music played by an orchestra with vocal calls interspersed:I imagine the dance-caller would have been stationed at the shorter horn “10” while “17” and “31” were aimed at the orchestra, with harder-to-record instruments clustered near the “17” horn. Virtually all of the technical inscriptions of the 0-prefix series can be interpreted plausibly in the terms I’ve just laid out, and if I’m right about what they mean, they offer rare insight into the strategies which Berliner’s recordists were adopting in an effort to capture increasingly complex subject matter to ever higher standards of perfection.
The established Berliner system of technical notations seems to have been carried over into Eldridge Johnson’s wax recording experiments that culminated in the manufacture of discs by the Victor Talking Machine Company. The Victor markings are often visible at three o’clock underneath the early “flush” paper labels of 1900-1902, and they’re similar in form to the Berliner markings I’ve just been describing.
In place of the single letters found on Berliner discs, we tend to find more complex alphanumeric sequences such as 1N3, 2N3, 4N3, 2S3½G, and 2S4G. Berliner’s sound-boxes had traced lines in a fatty wax surface on a zinc disc, but Johnson’s sound-boxes had to cut a groove in wax, so it stands to reason that Johnson would have needed to use different sound-boxes with different technical parameters, whatever they were. In the second part of the code, however, we find some of the same familiar numbers on Victor as on Berliner, often turning up in similar “fractional” layouts. The following examples are typical:
I’d read these as respectively documenting the use of a single “11” horn; a combination of two “17” horns; and a single “11” horn linked to a combination of two “17” horns. Elevens and seventeens are by far the most common numbers seen here on Victors, and they’re also among the most common horn numbers encountered on Berliners. The consistency isn’t surprising; after all, the horn is a part of the signal chain that, unlike sound-boxes, Johnson wouldn’t have had to change to accommodate his new wax recording process.
Alas, the “sunken” or “recessed” label type which Victor introduced for subsequent pressings obliterated any technical notations that may have been present on Victor masters, and even when present, the markings are often hard to make out underneath the paper labels, particularly when the labels are worn or scuffed. The bottom line is that Victor technical markings aren’t quite as accessible as Berliner technical markings are.
But there’s reason for hope on the Victor front. Several years ago, I got a brief look at some of the digital facsimile pages of Victor ledgers being used to compile the Encyclopedic Discography of Victor Recordings, since superseded by the Discography of American Historical Recordings (DAHR). In addition to the standard “discographic” data, I noticed there were columns dedicated to parameters such as “sound box” and “horn,” and that these contained notes resembling the technical markings visible on the discs themselves, along with occasional grids that seemed to represent floor layouts and the physical placement of performers. That was well before I’d begun trying to decipher Berliner technical markings, and I didn’t pay very close attention to specifics at the time. But judging from what the Berliner markings appear to be telling us, I suspect the technical columns in the Victor ledgers might provide stunning insights into the evolution of the strategies used by early recordists.
The use of pairs of recording horns for acoustic-era commercial sound recording isn’t much discussed in the literature, but scattered evidence of the practice can be found, including this photograph from the Library of Congress of Billy Jones in a Victor studio around 1920, cropped here from a larger image hosted on the Mainspring Press website:
One of the horns seen in this photograph was dedicated to picking up Jones’s voice, while the other was presumably dedicated to picking up an instrumental accompaniment from elsewhere in the room. In The Changing Sound of Music: Approaches to Studying Recorded Musical Performances, Daniel Leech-Wilkinson cites evidence of the use of similar two-horn setups for instrumental ensembles (online here, with a diagram from 1928 reproduced as Plate 3). The same strategy is apparently used for acoustic recording today as well, judging from contemporary wax cylinder recordist Shawn Borri’s recommendation on the Talking Machine Forum: “For duets and quartets, two recording horns should be hooked together with a horn copper adapter,and rubber tubing of 4″ long. The singers at one horn and the other towards the band or piano.” If I’m right, Berliner technical markings show that this two-horn approach was already in use at the beginning of the 0-prefix series in early 1899.
But Berliner’s apparent use of three horns starting in January 1900 starts drawing us into less well-charted territory. In that light, consider the following remarks made by Raymond Sooy in his “Memoirs of My Recording and Traveling Experiences for the Victor Talking Machine Company,” available online here:
In the acoustic recording we have used as many as twelve recording horns at one time with good results. This made it very difficult because the more horns used, the less volume you would get in the records, consequently, a very sensitive diaphragm had to be used for this purpose; then again, we were forced to use Stroh violins.
It seems that acoustic-era recordists didn’t stop with just three horns. So when did the jump to four horns occur—or five, or six? And what specific recordings involved the virtuosic use of twelve horns? What was the subject matter, and what do the recordings sound like? Did anyone ever try to use even more than twelve horns, even if this ended in failure? Are there cases where different takes of the “same” selection were made using different horn configurations or different numbers of horns? Can we hear a difference?
For now, I can’t answer those questions. Pushing this line of inquiry further into the Victor era would really require access to the ledger facsimiles. But I believe we can at least assess how things developed between March 1899 and May 1900, with the three-horn breakthrough falling sometime in January 1900. And that’s a good start.
However, I have to admit that I’ve seen one anomaly that throws a fly into the ointment. The technical inscription on Berliner 0489 nt—YD 17—appears to contain not one but two initial letters.
This recording couldn’t very well have been made with two sound-boxes labeled Y and D. I think I have the beginnings of an explanation, and one that’s consistent with the rest of my analysis, but it draws on evidence from technical notations found on discs recorded in 1897-1898. And that’s a subject for another post.