Dowland’s famous “tremolo” fantasia is familiar from many performances, starting with Julian Bream in the 1970s and continuing to be a favourite piece amongst modern lutenists.
This piece appears in the Collected Lute Music of John Dowland (London: Faber. 3rd edition, 1981; edited by Diana Poulton and Basil Lam; No.73). Nearly every modern performance I have heard has followed this text. The original piece is found in CUL Dd.9.33, a manuscript written c.1600 by Mathew Holmes. Here I consider the source text and the issues which arise when we look carefully at it. There is now an online facsimile of the manuscript, so you can examine it in minute detail:
Many pieces are found in more than one source and sometimes important information can be obtained by comparing sources. Unfortunately in this case this is the only known source.
The attribution to John Dowland is made on purely stylistic grounds but it seems to me to be plausible. What I want to do here is discuss the text and suggest some possible corrections, noting some of the errors in the Poulton and Lam edition (hereafter P73). The methods of analysis I suggest are applicable to all music, so I hope that my comments will be useful beyond the discussion of this one piece.
One important feature of my analysis is “barlines”, which are not musical barlines in the usual modern sense but – in lute tablature – are simply a way of keeping one’s place and have almost no musical significance, however it is usual for them to be regular (usually every two minims). P73 halves the note values of the original and introduces regular barlines, but my version (see PDF) reproduces the format of the original source (with a few exceptions, noted below). All bar references are to my version, not the P73 version. I apologise for the constant references to bar numbers, but it seems there is no other way to do it – you will need a copy of my PDF (preferably P73 as well) in front of you, otherwise what follows will be totally incomprehensible. References to individual notes are given by tablature letter and course number – so “e5” means “letter e on the 5th course”.
Peculiarities of rhythmic notation
The first thing to note about the original is the rhythmic notation – Holmes normally used the “gridiron” system where each note has a rhythm sign and groups of similar notes are beamed, very much as they are in modern mensural notation. But he also used a system – more common in printed books – where a group of notes of the same value have a rhythm sign just at the beginning of the group, which applies to all the following notes until a new sign appears. In Dd.9.33 he used a mixture of these methods, and also used an abbreviation whereby a dotted note was sometimes not given a stem at all, but indicated by a dot above the shorter note which follows the dotted note. This is a detail which was missed by the editors of P73, so some of their bars (1-2, 18-20) have incorrect rhythms. Here are the first two bars:
And here are bars 18-20:
I have corrected all eight cases where this occurs.
Points of imitation
The piece starts with a common formula for a “point of imitation”, then (after four entries) breaks off into a descending sequence in bar 5. However, the fourth entry is disguised at the beginning of bar 4 because the first note is an octave lower than expected (or not, because of octave stringing) and the third note is missing – the c1 at 4/3 should probably be c5. The rhythm signs (or lack of them) in bar 4 are a problem and I have offered a solution different from that of P73.
The next point of imitation comes in bars 14-18, where the first entry is bedevilled by incorrect rhythms, a displaced barline and an incorrect third note. In this case I have not only corrected the rhythm but moved the barline so that the first entry matches the following three entries. In bar 14 P73 destroys the first entry by failing to correct the third note (from a’ to g’) and then doubling the speed of the last four notes of the bar for no obvious reason. At the end of bar 18 P73 also introduces a G chord which is not in the source and destroys the final entry.
The final issue concerning a point is in the second half of bar 23 where an imitation of the previous two entries at a different pitch and inversion results in very inept counterpoint (implied parallel octaves). Here is the original text and four possible solutions:
Solution (a) is the “minimal intervention” approach – it changes only one note (c5 is added) but depends on having an octave string on the fifth course, and makes for a rather awkward bass line. Solution (b) adds the missing note but assumes the a4 is an error. Solution (c) adopts the little rising scale in the Tenor which is a feature of the previous two entries (bars 21-22). Solution (d) assumes that Holmes wrote e5 instead of e4. I leave it to you to decide!
Rhythms and barlines
Holmes’ barlines in this piece are mostly regular and mostly every four minims. Sometimes they provide important clues which help us to disentangle the rhythmic notation (or lack of it). The rhythm of the last two notes of bar 24 is puzzling – given the previous sequence, one would expect both notes to be crotchets (2 tails), but Holmes gives us quavers, making a 5-beat bar. In fact the previous few bars seem to start and finish “off the beat”, notwithstanding the fact that lute tablature barlines don’t necessarily have any implications of accent. I have adjusted the barlines so that bars 18-24 follow a regular pattern, without changing any of the notes.
Bars 35 to 37 are in triple time, according to Holmes’ barlines, yet in P73 the editors have deleted the third beat of 35 for no apparent reason. Whether 38 should also be a triple time bar (and the second beat repeated) is a question I leave to your judgement. Normal service in common time resumes in bar 39.
At the end of bar 39 the cadence is unsatisfactory as it stands and seems to imply a 9-8 progression where normally we would expect a 6-4, which I have provided as a suggestion.
I have added a barline at 46 for the sake of consistency and this also corresponds to a page break in the source where Holmes added some extra staff lines in an attempt to squeeze in the last two notes beyond the margin so it’s possible he didn’t have room (or didn’t see the need) to add a barline.
The tremolo pattern in bars 40-45 should probably be played thumb-index-middle-index as indicated in a tremolo exercise in the Board lute book (f.15v.):
Finally, the repeated patterns at the end of the piece give rise to some problems. For a copyist, there is always the danger of writing too many or not enough repetitions, and Holmes was not immune to this problem. I have given his text for bars 50-52 exactly as it stands in the source, so you can make up your own mind.
I’m not claiming to have created a definitive modern edition of this piece, but I hope to have shared some insights into the kinds of analysis which are essential for an intelligent performance of any piece which has come down to us in a less than perfect state – and with lute music, that’s just about all of them.
Digital images from Cambridge University Library are provided under license CC BY-NC 3.0.
Thanks to Francis Shepherd for his insights into “the bar 23 problem”.
Yoel Shamir has kindly pointed out a couple of errors and made some more suggestions about bar 23. I have added some comments of my own:
In bar 12, event 4 one might consider changing e3 to d3.
In bar 18, event 1 I might change a2 to f2; and at event 8 one could change a1 to e1.
In bar 21, event 1 the a6 should of course be a7 – this is an error in the original which I knew about but which somehow ended up uncorrected.
In bar 28, event 3, e2 could be changed to g1; and at event 9 one could change a2 to e2.
In bar 46, event 7 the c3 could be changed to a4.
In bar 23, we have two new suggestions, of which I prefer the second: