Schumann and the Cipher - Letters and Comments on Eric Sams's Essay

© The Musical Times, Oct. and Dec. 1965 (pp. 767-771 & 949)

There is no doubt at all that the cipher constructed by Mr Sams can be used to turn words into musical phrases. As he so aptly remarks, “it seemed that given patience the cipher could be trained to speak whole sentences even if in a halting and stilted way . . If this were a game it was a very agreeable one” (my italics). This high productivity of the cipher from Schumann's music can be explained in two ways: either that Schumann was using it, or that its rules allow so many options that it can be made to extract words from any musical material. Neither you, Sir, nor Mr Sams seems to have it clearly in mind that any “test” of the cipher should primarily seek to show that the second of these alternatives is unlikely. As Henry Moule used to say, it is a pity that musicologists are not made to learn a little elemen­tary statistics. Instead of asking readers to try the cipher on more works by Schumann, you might more profitably have asked them to see if they could build up an equally convincing case for the use of the cipher by, for example, Handel. You did try something of the kind, but mistakenly dismissed it as “irrelevant”.

    The other way to test Mr Sams' theory would be to ask him to lay down in advance of the test all the rules of his code, including all the ways in which he considers it possible that Schumann might have ex­tended or broken the rules. One would also require a strict definition of the kinds of musical themes or other combinations in which Mr Sams thinks one may look for words; and a list of German words, however many there might be, which he would regard as significant if found in a particular piece by Schumann. This piece, which would have to be one not previously examined in connection with the code, would then be thoroughly explored for coded words, using the rules but not allowing the slightest deviation from them apart from those specified in advance. Meanwhile a statistician would calculate the number of words he would expect to emerge from the piece by chance alone.

    This piece of research is not likely to be undertaken, but I have described it to show how very far Mr Sams is from having given his cipher any real test. He gives us the cipher (Ex 10) along with the rules for its application, and then proceeds to apply it to Schumann's music. But we find that in every case but one he has to allow some special extension of the cipher or breach of the rules or other “fiddle”, before the notes can be made Io yield the desired significant words. And this despite the fact that the cipher already allows a choice of at least six letters for any note, and a further choice of seven “modes” (produced by shifting the cipher along the scale).


    Ex 11a: the word Clara carries no weight, since the cipher has been constructed so that that phrase will yield the word Clara. The rest of the message has two letters omitted. (Sams blames Schumann for this!) Moreover, in both this and Ex 11b a new device has been introduced: transposition of the music, instead of shifting the mode of the cipher as originally described. Why? Because shifting the mode does not yield the desired message. The second note, E in Ex 11a, can be made to yield letter O by calling it Fb; by the other method, a double flat would have been needed, and even Mr Sams has not admitted this degree of “flexibility”. Ex 11b: here it is claimed that the slurred notes have been picked out for the message, but in fact another note (the fourth) is also slurred in the Complete Edition.

    Ex 12: Instead of musically phrases, we here enter a new domain: that of picking out the accent signs, even if separated by 16 bars, and looking for mes­sages in the series of accented notes. But even so, in the last bar of the example Mr Sams has had to ignore the accented bass note corresponding to those he has used in the previous bars, and jump suddenly to the treble; and the word that emerges seems to have no significance.

    Ex 13: One note from the introduction is arbi­trarily joined lo the real musical phrase; one letter is missed out of a word; and again, the words seem to have little significance. (The identification of the alto phrase with Clara is another matter, but this, even if true, gives no support to the cipher theory.)

    Ex 14: A word, and a letter, left out; and the succession of notes chosen for decoding wanders between bass and treble without any musical justification. If this is allowed, then one could pick and choose from bass or treble at will: another powerful extension of the rules.

    Ex 15 and 16: Here Mr Sams has had to change the cipher itself.

    Ex 17: Further wanderings between the hands, and a brand new extension: the anagram. Schumann did not use the phrase (ex A) which the code


makes out of the name Paganini: he jumbled it up instead, we are told (ex B).

    Ex 18: the only one that strictly follows the rules. But the result is a phrase of only four notes and of doubtful musical importance.

    Ex 19a and b: Accent signs again. I have not seen the piano duet version which Mr Sams found he had Io use, but in the orchestral score the par­ticular notes chosen to yield the names Hermann and Dorothea are neither the only ones accented nor the ones accented in a particular orchestral part. Are we asked to believe that Schumann first wrote the orchestral version, then, finding that the music he had already written could be made to function as a coded representation of the names, placed the accents accordingly in his piano-duet arrangement ?

    Ex 20: This phrase does not occur at all in Op 23. There are phrases similar to it, but these do not yield the word Eduard.

    Ex 21: Another extension is introduced here-the use of a phrase that does not actually occur in the music but which, in Mr Sams' view, is the “source of basic motivic material”.

    Ex 22: Two letters omitted from the word; two notes tacked on to the beginning of the musical phrase, and the last two notes of the phrase replaced by two others.


    You have confused the issue, I think, by saying “Schumann was a composer: the fact that he occasionally "cheated" to produce a better musical result again seems to me a strong point in the cipher's favour'. It would indeed be useful to be allowed to cheat if we were using the cipher for musical composition. But we are not. We are testing a hypothesis that the cipher was used by Schumann. To cheat in this situation is fatal.

    I do not by any means think it impossible that Schumann used ciphers of some kind, for his own secret pleasure, or possibly as a starting-point for thematic invention-in the same way that some composers have used rows or random numbers. But Mr Sams' cipher is only one of thousands of possibilities; and the evidence for its use by Schumann evaporates when one studies it closely

Clare College, Cambridge                                                                        Nicholas Temperley


Eric Sams replies:


It is good of Dr Temperley to have investigated the cipher in such detail. And certainly I cannot com­plain if he is sceptical; I asked for comments and expected a fair share of healthy scepticism. I am sorry therefore to seem less than grateful. But what sought was help in testing and judging the evidence as a whole; whereas what we get from Dr Temperley is something rather different.

    His conclusion is that Schumann may well have used a composition-cipher, and used one flexibly (“cheated” or “fiddled”). Yet he insists that the right test is the rigid one appropriate to a communi­cation-cipher, “not allowing the slightest deviation”. In other words, the test is specially designed to be irrelevant: a novel principle of experimentation. However it also appears that no tests of any kind actually need to be made by Dr Temperley or any­one else, nor need any statistical evidence be ad­duced, or any evidence of any kind; nor need he look too closely at my evidence (eg Ex 18-22) or even the basic facts (the cipher gives at least three, not six, letters per note). He just assumes that I must be faking it all; and he writes, and comments, on that assumption: Why he makes it he does not say, and I cannot tell. It is certainly arduous to argue with, and to that extent effective. But is it quite fair or quite reasonable? If Schumann, as he says, might well have made up and used a flexible cipher, why should it not be the one I describe? Let us consider the detailed points on that counter­assumption, and see how it works.

    Ex 11: the word Clara would be enciphered just like any other word, with just as much value; the manuscript omits no letters; the music would have been composed in B minor and transposed to fit the Dichterliebe key-scheme; to 5chumann's ears Fb would not be “called” E but would be E. I don't see the point of noting that another slur was put on his cipher long after he was dead. The idea of enciphering after this pattern (Ex 11b) could have been taken over from Kreisleriana No 1, which the music much resembles; a test on that might be reasonable.

    Ex 12: but why not use accents? The wide separation seems to make the result more striking, not less. Schumann was not obliged to use the bass accent (perhaps it was put in later for consistency's sake, or perhaps the ciphering continues). Still less was Schumann obliged to make a phrase that would later be thought interesting by Dr Temperley.

    Ex 13: Why should Schumann begin with what someone else thinks is the musical phrase? How­ever, perhaps he did; the cipher is phonetic in more ways than one, and ww know from Carnaval that Schumann was quite capable of thinking of S as Es, and so perhaps of R as Er. Both are so pro­nounced; further, such usages are described in cipher manuals (see 2 below). Here Dr Temperley seems to concede that there may be something in the “Clara” theme; but if so, there may be something in the cipher, since that was its starting point.

    Ex 14: The omissions are commented on in my article, which specifically makes the point that Schumann would be using a cipher for his own pur­poses, not for those of later testing or commentary. The question here and elsewhere is not whether the thing was done well or interestingly, but whether it was done at all.

    Ex 15, 16: but why not assume that it was Schumann who changed ii, in the way and for the reasons suggested?

    Ex 17: but why should not Schumann have done this?

    Ex 18: but musical importance, whatever that means, begs the question; what happened to those “elementary statistics”?

    Ex 19: I deal with this in more detail later. Why couldn't Schumann have written his cipher accents in a draft piano version?

    Ex 20: certainly a phrase giving Eduard occurs in Op 23, as 1 quote it; I wrote it twice, in the treble clef, to make the idea (as I thought) clear.

    Ex 21: certainly these phrases occur in the music. The notes of (a) occur in due order separately in bars 1, 9, etc, and together in bar 5 etc, while (b) (as corrected in the footnote) is the idea of the B major section, the right-hand melody in the fourth bar of which actually spells out “Macbeth” note by note.

    Ex 22: This is misrepresentation; I nowhere say or assume that Schumann had to use his cipher to spell out a name, letter by letter, or to do anything. Why should he? I really do not see how this point or any of the others can be supposed to go anywhere near, let alone touch, much less damage, the cipher hypothesis.


    So much for irrelevant test and assumed con­clusion. Now for the neglected evidence. In a process of inductive inference of the kind described in my article, evidence is cumulative, according to the rule that to find the probability of getting all of several independent results we multiply together the chances of getting each one separately.

How probable is it that any or all of the following results could be claimed at all, unless on the cipher hypothesis? (References are to the August MT, which contains all the essential data.)

    1) From a simple working hypothesis (that the word, “Clara” was enciphered in a given way) it was possible to derive an arrangement in almost perfect alphabetical sequence, which looked like, and yielded, a viable cipher system (Ex 6).

    2) This system combined (a) a note-for-letter musical substitution cipher; (b) added sharps and flats; (c) a variable setting; (d) a three-line alpha­betical arrangement; (e) an alphabet of 24 letters; (f ) abbreviations such as R for ER and F for PH; (g) the extraordinary usage of Q for CH and X for SCH. I had never previously come across any of these devices. But I subsequently found each of the first six clearly and separately stated, and the last as clearly implied in a German book on cryptography (Klüber 1809) which at the material time was quite possibly known to Schumann.

    3) The first cipher table (Ex 6) was bare of interest (Ex 7, 8), while a slightly revised table (Ex 10), differing in only five letters, was - and is - full of it (Ex 11-22 et al).

    4) Ex 10 produced al its first trial, and within twenty minutes, a coherent text of 19 (or using Schumann's MS, thirty-three) consecutive symbols.

   5) Ex 10 worked at once on passages chosen for what had long seemed their cryptic qualities ­while numerous other passages, chosen at random, were without result.

    6) Ex 10 fails to work on the early music (eg Op 2 or Op 8). But these responded to a different table (the letters A-X written out in order in three rows of eight), which in turn did not work on the later music.

    7) Ex 10 differed from this basic table (top of p.589) only in its first row of letters; the change was precisely that required to turn the basic table into one giving a stepwise melody for the name “Clara” - a melody already described, and linked with Clara, by others.




With this general evidence go the particular ex­amples of correspondence between the music and an external name or word. Each of these gives a further quota of probability to the cumulative total. Take for example Ex 19, where the odds are most readily quantifiable. This shows that successive accented notes in the treble clef of the piano duet version of Hermann and Dorothea spell out those names in the first and second subjects-in accord­ance with a cipher system independently evolved and described. (Incidentally, in my copy, Peters No 2358, arr Horn, the decipherment disappointingly ended at bar 79 with the “e” of “Dorothea”; but the first edition, Schumann's own arrangement, has not only all the accents needed up to that point, but also, in bar 84, the extra one needed to add the “a” and complete the name.)

    Ex 10 et seq shows that any given note must correspond either to one, to two, or to three three­letter columns, out of eight such columns. So the chances that any given note will correspond, just by coincidence, to the column containing any given letter will be one, or two or three out of eight, as the case may be. Given the probability rule already mentioned, the question becomes more one of calculation than speculation.

    There may be room for argument about the exact way in which this rule is to be applied. But it is not easy to see any ground for disputing its general validity or its plain conclusion, namely that this example, which is itself only one fact among many very singular facts, could occur by chance only at odds of many millions to one, even when variant settings are taken into account. Can this be just coincidence? But if it isn't coincidence then - if I understand Dr Temperley's alternatives aright­ - must be the cipher I describe. Well, I don't mind conceding that he may have a point there.


                                                                                                                              Eric Sams




Although a good deal of Mr Sams's article seems to go rather far in finding words to fit, there is one little cipher which immediately rings true. This is the use of the notes EHE in the wonderful setting of Eichendorff, Mondnacht. Here the poet suggests a sort of “marriage” between heaven and earth, while the composer seizes on the idea of his own marriage to Clara.

    What is so extraordinary is that, however arti­ficial may be the means, the result can be such a perfect masterpiece. The EHE motive forms the second half of a seemingly strophic idea. The first half, so subtly different each time and so memorable because of its clashing E# and Eb, is an example of another feature of the composer. Highly inspired, he often taps the mind of someone else, frequently Beethoven, but without suggesting “cribbing” - in fact a psychic personality.

    In this case, the unexpected source is Haydn's Nelson Mass, from the “Et Resurrexit” at the words “et vitam venture”. The marking vivace to this move­ment makes it sound very different from the lan­guorous moonlight of Schumann's song and yet the harmonic progression, not a common one, is the same in both, kicking off unexpectedly on the sharp side from a dominant. Is it fanciful to relate the “Life Everlasting” of Haydn with the “Heaven which kisses the Earth” of Schumann?

University College of Wales                                                                                            Ian Parrott





I find Mr Sams entirely convincing. Not only does his discovery remove many of the perplexities surrounding Schumann's music, but it can also be used to throw some light upon Brahms's relations to the Schumanns. it is well known that Brahms's feelings towards Clara went beyond those of simple friendship, but a natural shyness towards women, together with a high regard for Clara's husband, made it impossible for him to declare his 'love openly, even after Robert's death.

    That Brahms's passion for Clara found ex­pression in his music in a general way is obvious Io all who know the music well. What is now so striking is the way that the Clara motif frequently emerges in places where its significance is un­mistakable. The songs quite naturally provide many examples, among which one might mention Op 46 No 1 at the words “schmerzlichen Erguss” (painful effusion); Op 72 No 3 at “in Schmerzen schlief der Wiederhall” (in grief the echo-of his love-sleeps); Op 85 No 16 at “Ich sass zu deinen Füssen” (I sat at your feet); and Op 71 No 2 at “Vertrauter meiner Schmerzen” (sharer of my sorrow). The Clara motif is very prominent in the Liebeslieder waltzes too, and provides a great deal of the material for the Clarinet Quintet - one of the most passionate pieces of music outside Tristan.

     We can easily surmise what happened. On one of his visits to the Schumanns, Brahms must have chanced upon the cipher while Robert was out of the room. Realizing its importance, he hastily made a copy of it, and had time to replace the original before Robert returned. He was now able to confess his love for Clara more explicitly than had previously been possible, but of course he still had to be careful not to betray himself to Robert. It is most significant that the Clara motif does not emerge from his music with any clarity until after Schu­mann's death in 1856.

     Like Schumann, Brahms made use of the cipher to invent new melodies. His own Christian name, for example, gave him the idea for some of his best­-known songs: The Blacksmith, Feldeinsamkeit, and To a Nightingale (see Ex 1). Any doubts we may still have can soon be displaced if we ask our­selves one obvious question: “Where, if anywhere, would Brahms be likely to declare his passion beyond any possibility of misunderstanding?” The answer is surely in the first symphony - the work whose composition had cost him more pains than any other, and one to which he attached tremendous importance.

    We do not have to look far. We can leave aside the slow introduction, since “Frau Schumann's letters and diaries show that the first movement originally had no introduction” (Tovey, Essays in Musical Analysis I, p.84). This means that Brahms's original intention was to open the work with a phrase in which his dedication to Clara is made absolutely clear (Ex 2). It is not, however, until a few bars later, when this phrase is combined with the first subject proper, that the full content of Brahms's message is laid open. Now, at last, after so many years of concealment or, at best, obscure hinting, he makes his love known to the only person in the world who can read his meaning­ Clara Schumann (Ex 3), “Lieb' ich mein Herz, Clara” he says; I love Clara, my Heart. Like Schumann in Mr Sams's Ex 11 (a), Brahms was unable to complete the word mein, but we now know why the earliest sketches for the work have an A fiat in place of the D in the fourth bar.

    Brahms's friends came near to breaking the code when they forced him to admit that he had adopted the notes F-A-F as a kind of motto in his compositions. When pressed to reveal the significance of this motto he had to think quickly, and invented the phrase “Frei aber froh” to satisfy their curiosity. These quite meaningless words have been accepted ever since. Now, thanks to Mr Sams's researches, we can reveal the truth of Brahms's obsessive motto. When he wrote the notes F-A-F, what he intended was nothing less than a proposal of marriage. “EHE!” he spells out, again and again. But the only one who could interpret his meaning would not reply.

Cardiff College of Music & Drama                                                                                   Malcolm Boyd




Eric Sams's article on Schumann's use of ciphers seems quite convincing and certainly makes for fascinating reading. I strongly suspect that a goodly number of composers have on occasion done similar things, which have not yet been ferreted out. lf the secret codes are too obscure, the chances are the composers themselves will remain the only ones to know about them. Schumann, apparently, was sporting enough to leave some clues to put Mr Sams on the scent. At one time, I might add, I “signed” my own compositions with a rather obvious cipher, which I managed (or so I thought) to conceal cunningly.

    In one work the “signature” annoyed me every time the passage was performed, so I desisted from then on.           

Asolo, Treviso