I have had several goes at this article, only to find more complexities and inconsistencies each time. It began with a narrow focus: Malcolm MacDonald, in The Symphonies of Havergal Brian, vol. 3 p. 64, and again in his inlay notes for the Marco Polo recording, writes unequivocally that the last two movements of Symphony 32 play without a break. The evidence for this seemed to me less than clear cut and, I thought, arguably outweighed by evidence pointing the opposite way—unless there was some obiter dictum by Brian on the subject which I did not know about.
It arose for me because Brian’s last symphony is one I have been gradually setting on computer, having had to make a start, in a frantic rush ahead of the California performance last November, with the four parts which had gone missing from UMP’s hire set. The computer-‘engraving’ of a score, especially for an amateur enthusiast with a pension, is or should be rather more than just keying notes and markings into a computer, and enjoys huge time advantages over sitting at a library table with a stack of manuscript scores and time racing by. When time is not a problem, one feels obliged to study almost every note in a critical way, fretting about where the notehead is placed (often ambiguous with the speed at which Brian wrote) and about whether a slight variation between several instruments playing what looks like the same sequence of notes is a mistake or has some subtle purpose. For example, Brian’s idiosyncratic use of hairpins, rarely aligned with any care and often left to be inferred, is one of the ongoing headaches for the Scores subcommittee and deserves its own article once some tentative rules of interpretation emerge.
So the narrow focus is on the junction of movements three and four of Symphony 32, but the study sent me out into those other symphonies for which I have the manuscript in photocopy: The Gothic (Part 1), and Symphonies 2, 29 and 32 itself. (I also have nos 6 and 16, but they are one movement symphonies anyway.)
The problem arises from the manuscript full score. The first two
movements end with a double barline, no cautionary time-signature
before the page-turn
(whereas Brian normally does put a cautionary time signature if it arises at a page-turn within a movement), and no verbal instruction such as attacca; and the following movement is headed by a bold Roman II or III. All perfectly standard. But p. 35, where the third movement ends, shows only a single barline at the end of the system, and p. 36 does not have a ‘IV’ at the top. Except for the absence of a cautionary time-signature there is nothing to to indicate
that they are not one movement. The presence or absence of a verbal instruction is the obvious point to look at first, but it proves to give little help. In The Gothic (Part 1) Brian writes attacca at the end of the first movement, even though it would create a most uncomfortable jolt to progress straight from the concluding ff chord on the first beat of a brisk 4/4 (with no closing rests) to a 5/4 ‘Very Slow and Solemn’ played pp. If this is attacca, it is certainly not subito. Yet at the junction of the second and third movement there is no attacca at all, even though the transition is seamless: the reliable inference of
continuity comes instead from the held note of the bass clarinet.
Symphony 2 has the Roman numeral at the top of each movement, but
the first and third movements conclude with the instruction
attacca, with effects
which MacDonald acknowledges in vol. 1 of The Symphonies of Havergal Brian. On the other hand, Symphony 29 compounds the confusion: each new
movement has the bold Roman numeral, but the end of the first has no double barline, the end of the third actually has a cautionary time-signature change, and at all three junctions we see an attacca, as if the whole work were to be played without a break; the principal evidence for a smooth join of the first two movements, and of the latter two, but for a slight pause between the second and third, is to be found in the instrumental writing.
In Symphony 32 the eye is admittedly drawn to a word in the right margin at the end of the third movement, half way down the page, that looks a bit like Segue (see illustration). There is an ambiguity in the last letter— ‘a’ or ‘e’?—where Brian’s hand would have been tilting off the right edge of the paper stack; the word begins with ‘Se-’, and the fifth letter is either ‘g’ or ‘z’, but there is no ambiguity about the middle letters: look at them closely—I’m afraid my eyes needed a powerful magnifying glass to be sure—and they are certainly not ‘-gu-’ because the letter with the descender comes after the ‘n’ or ‘u’, whichever it is; but you can reasonably read ‘- nz-’. This, then, is no more than an instruction to trumpet 3 to remove the mute, with which it has just played a held note, in readiness for the next movement where it has no mute when next on duty. Senza, much more clear, also appears against trumpets 1 and 2, and more fully as Senza Sord. against horn 4. I think there need be no doubt that the word is not Segue.
What of other notations relating to the transition between
movements? MacDonald (ibid p. 74) points out that the third
movement has one and a third beats’ rest at the end, an observation
which turns out to be crucial in the performances on record. Head
(1971) handles the transition in just this way, with those
rest-values and then, in effect, an attacca so that there is a
literally measured short pause but not what one might call
‘coughing time’; no real opportunity for the mind to shift gears in
readiness for a new movement, instead just time to catch a quick
breath before being hurried onward into a new but connected
section, albeit at a different tempo and with a different rhythm.
This is what one would expect, with hindsight, from the footnote
anecdote on p. 64 of MacDonald (about critics who were guided, not
by what they heard if indeed they bothered to stay for the
performance at all which not all ‘critics’ seem to do, but by a
mistake in the programme which referred to three movements).
Fredman (1978) allows a full normal gap between the movements.
Leaper (Marco Polo 1992), like Head, counts the beats but allows no
unwritten pause between pages 35 and 36. Russell (2005) fully
separates the two movements. The experience of listening
attentively to these two approaches does not, at least not for me,
establish a clear feeling of one being correct and the other not,
though I admit to a disagreeable sense of being jostled along by
Head and Leaper. But then again, Brian writes in the rests to fill
out the last
bar of each of these four movements, ie, even at the end of the work. So we can infer nothing useful from there being rests at the end of the third movement.
Some incidental support for the ‘without a break’ view comes with MacDonald’s comment about the first two movements, and then again the latter two, being in a sense partners in two large contrasting halves: ‘the first brooding, troubled and searching, … the second defiantly energetic and optimistic’ (ibid p. 64). The support would have been greater if the junction between movements one and two had been notated as between three and four; as it is the observation remains valid, as no doubt intended, in terms of overall structural effect but does not help in the narrow point at issue. Symphony 32’s short score (the short scores can often be most helpful in resolving problems of interpreting the full score) casts real doubt on Brian’s intentions for the end of the third movement. He unambiguously closes it with a double barline and heads the next with a big, bold IV. Did he change his mind for the full score? Or did he, by no means for the first time (though for the last, of course), lose track of what he was doing at the end of the page?
It seems to have been his working habit to set out each new page after completing the previous page, first sorting out how to make economical use of the staves by, for example, squeezing three percussion instruments onto one stave in order to gain the stave needed for an extra system; and then to complete the notation for that page. I imagine him about to turn in for the night when approaching the end of this page, or perhaps being summoned to tea, probably wishing to get to the end of the page first, returning to the job later with a fresh new page in front of him but now with a certain conceptual discontinuity in the sequence he is putting to paper.
This sort of thing often shows up in uncancelled technique indications (such as arco after a pizz. passage). There is a point early in Symphony 2—track 2, from 2’53” of the Rowe on Marco Polo—where Brian’s initial layout forgot that the violas had already gone into the treble clef on the previous page; and although for the Musica Viva edition he spotted this one and did squeeze an almost-invisible treble clef just after the regular alto clef, any copyist can be forgiven for missing it, leaving the violas to play a seventh below what they should be doing, and the result still sets my remaining teeth on edge. There is a prime example in Symphony 16 towards the end, at the turn of manuscript pages 26 to 27, which cannot be played as it is written, where there are several different ways of making it playable, and which also deserves an article all to itself.
Of course, the fact that a page-turn seems to be for Brian the
analogue of an accident black spot—and his penmanship was at
dangerously high speed as
his family recall—does not of itself mean that he made a mistake here, merely that we cannot wholly rely on the absence of double barline and Roman numeral. It is necessary to look for other indicators such as the absence of a cautionary time-signature (though even this practice is not consistent).
Whichever way you look at it, Brian missed out something from his full score at the end of the third movement and the beginning of the fourth. In the absence of any written instruction to play on without a break, or of any significance in his completing the closing bar of the third movement with rests, the evidence of the short score and the absence of the cautionary time-change seem to me to carry more weight than the absence of a Roman numeral at the head of the fourth movement and the absence of a double barline at the end of the third. So unless Brian has indicated elsewhere that these two movements are to be played without a break, I submit that there is a modest balance of evidence against so doing. Before his California performance last November, Chris Russell raised this issue with me and, while I doubt if my untutored opinion would have seriously swayed his choice, I believe he made the right one.
© Alan Marshall 2006