It is still widely believed that to play Dowland one needs an eight-course lute tuned in unisons throughout. It is also commonly believed that English musicians preferred unison stringing while continental lutenists used octaves. These beliefs seem remarkably persistent in spite of the fact that the evidence which contradicts them comes from Dowland himself and has been available for a long time. It therefore seems like a good time to set the record straight on these points, but also to look more closely at the requirements of his music and what kind of lutes he might have played.
Number of courses
If we look at the modern edition of Dowland’s solo lute music
(Poulton & Lam, 1982) we find 41 pieces for six courses, 43 for
seven (37 with the seventh at D and six at F), three for eight courses
and five for nine. This gives only a rough idea, because these pieces
are only the versions chosen by the editors – different versions
of the pieces in some MSS may require more or fewer courses and it is
mostly not known whether any of the versions are actually those of
Dowland himself.
If we look at Dowland’s printed works, we find:
1597 The First Booke of Songes or Ayres: seven courses
1600 The Second Booke of Songs or Ayres: seven courses
1603 The Third and Last Booke of Songs or Aires: seven courses, also a bass lute with seven courses
1604 Lachrimae or Seaven Teares: nine courses
1610 A Musicall Banquet (Robert Dowland): up to nine courses
1610 Varietie of lute-lessons (Robert Dowland): six to nine courses
1612 A Pilgrimes Solace: eight courses
Another way to approach the question is to look at the history of the
increase in the number of courses in general. The problem with this is
that for most of Dowland’s solo lute pieces we have no secure
chronology anyway. However, from the approximate dates of the MSS in
which they are first found, it seems likely that most of these pieces
were composed before 1600. We need to consider developments on the
continent as well, because Dowland was in France from 1580 to 1583, in
Germany and Italy from about 1594 to 1596 and in Denmark from 1598 to
1606. While seven-course lutes were known as early as the late 15th
century, most 16th century publications of lute music are for a
six-course lute. The earliest published pieces for seven courses are
found in the books of Melchior Newsidler (1574), Barbetta (1582) and
Adriaensen (1584). Terzi’s first book (1593) is for seven
courses, his second book (1599) for eight. Molinaro’s book of
1599 is for eight courses. Publications for nine-course lute start with
Francisque (1600), Besard (1603), and John Dowland (1604).
Right hand technique
We don’t know exactly when Dowland started to play the lute, but
if the career of Robert Johnson is anything to go by it is likely that
he was apprenticed to a master when he was about seven, which would
make it about 1570. This is about the same time as Adrian le
Roy’s instructions which were published in 1568 and subsequently
paraphrased by Barley in his A new Booke of Tabliture in 1596. Dowland
would also have been familiar with French practice from his visit to
Paris as assistant to the French ambassador in 1580-1583. So for
roughly the first thirty years of his playing career Dowland would have
played a six- or seven-course lute probably using what we now call
“thumb-inside” technique. As late as 1603 (in The Schoole
of Musicke) Thomas Robinson is still recommending thumb-inside, though
he is using a seven-course lute. With the increase in number of courses
in the last two decades of the 16th century more and more players
switched to “thumb-outside”, as recommended by Besard in
1603. Dowland is mentioned in the Stobaeus MS of about 1619 as being
one of the famous lutenists (along with Huwet, Lorenzino, and Bocquet)
who changed from thumb-inside to thumb-outside during the course of his
career.
Type of Lute
There is one contemporary reference to a “Venice” lute in
connection with Dowland: epigram 99 of Henry Peacham's Thalia's
Banquet, (1620, sig. C8v.) is addressed to "Maister Doctor Dowland":
Your word, Hinc illae lachrimae, beneath,
A Venice Lute within a laurell wreath.
Records of imports show that Venetian lutes fetched much higher prices than others (see Spring, pp.68-69). Besard (1617) and Fuhrmann (1615) depict ten-course lutes with large multirib bodies in the style of the German makers working in Northern Italy in the late 16th century.
Octave stringing
Dowland expresses a preference for unison stringing in 1610, and this
has often been taken as a justification for playing his music with
lutes strung in unisons throughout. It has also commonly been assumed
that unison stringing was a feature of English, rather than Continental
practice. In fact, Dowland himself contradicts both of these modern
views:
Secondly, for on your Bases, in that place which you call the
sixt string, or r ut, these Bases must be both of one bignes, yet it
hath been a generall custome (although not so much used any where as
here in England) to set a small and a great string together, but
amongst learned Musitions that custome is left, as irregular to the
rules of Musicke.
To take the second point first, he says that octaves were "not so much
used any where as here in England” – and he had first-hand
experience of what continental lute players were doing. In fact it is
the Neapolitan lutenist Fabritio Dentice who is credited with
introducing the practice of unison stringing. To return to the first
point, Dowland’s remarks in 1610 concern specifically the sixth
course on a nine-course lute and say nothing about the lower courses
(“accessories” as he calls them) which were almost
certainly always tuned in octaves. In any case, Dowland says that an
octave sixth course was common practice in England – of course we
cannot be sure of the timescale, but it seems likely that it persisted
until close to 1610. In fact, whether or not it ever became widespread
is debatable, because octaves from the sixth course downwards were
common on “baroque” lutes of all kinds from 10 to 13
courses, where the new tunings require a thinner sixth course than in
the old tuning and would therefore have had less need of an octave.
Returning to Dowland’s solo lute music, if (as seems likely)
nearly all of it dates from before 1600, then it would all have been
composed for a lute with an octave on (at least) the sixth course. In
fact, internal evidence from some Dowland pieces (the best example
probably being K.Darcy’s Galliard) and others from the 1590s
suggests octaves on the fifth and even the fourth course. It may be
that Barley’s instructions (1596), describing a six-course lute
with octaves on courses four to six, though taken from an earlier
treatise by Le Roy, still reflected common practice in England in the
1590s.
String types
Dowland's comments on strings provide us with a large proportion of the
tantalizingly little information we have about strings. The full text
of these comments can be found here.
In Dowland's time all lute strings were made from gut and there were no
metal-wound strings. Compared to the stringing which has mostly been
used on lutes in modern times (nylon trebles and wound basses), all-gut
stringing probably resulted in a brighter, more sustained treble sound
and a less sustained, darker bass sound.
Tension
Dowland recommends selecting strings according to the size of the lute:
...proportionably size your strings, appointing for the bigger Lute the greater strings, and for the lesser Lute the smaller strings
The use of bigger strings for bigger lutes also implies more string tension, if lutes were sized in strict proportions according to pitch, which is consistent with the relative sizes of surviving lutes. He also emphasizes the importance of grading the sizes of strings to achieve an even feel across the strings:
...these double Bases likewise must neither be stretched too hard, nor too weake, but that they may according to your feeling in striking with your Thombe and finger equally counterpoise the Trebles
...Thus as the sounds increase in height, so the strings must decrease in greatnesse: Likewise by the contrary, for those Accessories, which are the seaventh, eight and ninth string, &c. keeping the former counterpoise, as if they were equall things waighed in an even Ballance.
Double top string
The use of a double first is rare on modern lutes, but Dowland used it
in 1610 and it is common in surviving lutes of the period. Robinson
(1603) and Mace (1676) also used a double first.
Frets
...all the Lutes which I can remember used eight frets, and so ended at the Semitonium cum Diapente.
But yet as Plautus saith, Nature thirsting after knowledge, is
alwayes desirous to invent and seeke more, by the wittie conceit (which
I have seene, and not altogether to be disalowed) of our most famous
countriman M. Mathias Mason Lutenist, and one of the Groomes of his
Majesties most honourable Privie Chamber, (as it hath ben told me,)
invented three frets more, the which were made of wood, and glued upon
the belly, and from thence about some few yeeres after, by the French
Nation, the neckes of the lutes were lengthned, and thereby increased
two frets more, so as all those Lutes which are most received and
disired, are of tenne frets.
Mathias Mason was appointed as court lutenist in 1580, and promoted to
the Privy Chamber in 1603; in 1609 his name disappears from the
accounts so he probably died in that year. Notice that Dowland says
"three frets more", not four, as might be expected. This is probably
because these frets corresponded to our 9th, 10th and 12th, missing out
the 11th - the use of "m" to mean 12th fret (in the Marsh Lute Book,
p.153) bears out this interpretation. Dowland nevertheless implies that
this is a bit of newfangleness of which he might approve but in which
he does not personally indulge ("...I have seene, and not altogether to
be disalowed"). In fact he uses (the position of) the 11th fret (King
of Denmark's Galliard, P40, bar 25) and even the 14th fret (Mr
Langton's Galliard, P33, bar 53) - but presumably played all notes
above the 8th fret (early in his career) and 10th fret (later in his
career) without wooden frets. Body frets were used by some, however, as
evidenced by Robinson (1603) who, explaining the use of letters to
label the frets in tablature, says "...until you come to i which is the
last fret about the neck of the Lute, but you may glue on more frettes
in fit place and space (untill you come to n)"
Dowland is the only written source of information we have about fret
thickness (though some paintings, notably Holbein's The Ambassadors,
clearly show quite thin double frets). He tells us to use a fourth
course string for the first and second frets, a third course string for
the third and fourth frets, a second course string for the fifth and
sixth frets, and first course strings for all the rest. We should
remember that when he talks about string sizes he is not talking about
precise measurements, only groups of sizes (as recently as the early
20th century it was usual to classify violin strings as to whether they
were made of three or four guts, actual diameter could vary
considerably within those categories) - so Dowland's fretting could be
continuously graded rather than stepped. The most important implication
is that Dowland used very thin frets: probably starting at about .80mm
decreasing to less than .45mm. These frets were almost certainly
double. This contrasts with the modern practice of using very thick
single frets.
String length and pitch
I think that Dowland's statement (above) that the French lengthened the
necks of the lutes and thereby gained an extra two (tied) frets is to
be taken at face value (the alternative, of course, is that new lutes
were made with smaller bodies and relatively longer necks, but it seems
much more likely that existing lutes were converted). If we take as a
starting point a lute of about 59cm string length and add two frets to
the neck we would get a string length of about 66cm. Assuming a treble
string tuned close to breaking point in both cases, we would expect a
drop in pitch of about a tone. If, however, we were simultaneously
converting the lute from six courses to nine, the extra width of the
neck required by the extra courses means that the body will be
shortened (in this case probably by about 1.4cm; the exact figure would
depend on the shape of the body, but I am assuming a Venetian/Paduan
style lute), and the string length might then be only about 63cm.
Nevertheless, the drop in pitch would still be at least a semitone.
A pitch around a tone below modern also fits with Ian Harwood's theory
(Harwood, 1981) that there were two pitches in use in England around
1600 about a fourth or a fifth apart, our pitch of about a'=392 being
the lower of these two pitches.
Other evidence for a possible drop in pitch comes from tuning
instructions. Robinson (1603), in common with most previous authors,
recommends tuning the first course as high as possible (“so high
as you dare venter for breaking”) and tuning the other strings
using the first course as a reference. On the other hand, Dowland
(1610) says:
...first set on your Trebles, which must be strayned neither too stiffe nor too slacke, but of such a reasonable height that they may deliver a pleasant sound, and also (as Musitions call it) play too and fro after the strokes thereon.
Another factor in trying to deduce pitch is the use of a double first string. There is a limit to how thin a string can be made (modern best guess is about .42mm) and Dowland himself warns us against using treble strings which are too thin:
...take one of the knots in your hand, but let it not be too small, for those give no sound, besides they will be either rotten for lacke of substance, or extreame false.
If we take a modern lute of 59cm string length and assume a maximum working tension for a single treble string of about 40 Newtons, the pitch of our hypothetical .42mm string will be approximately g' at modern pitch. If we take a larger lute, say our hypothetical converted lute with a ten-fret neck and a string length of 63cm, to maintain the same tension the pitch would have to go down by about one and a half semitones. I think it unlikely that a double top string would be used at a total tension of 80N for the course, so use of a double first would push the pitch down still further. The issue here is not the pitch at which a string might break, but the kind of tension which might be acceptable to a player.
Conclusions
Bearing in mind the duration of Dowland's lute playing career, from
c.1570 to 1626, and the changes in instruments, strings, and playing
techniques which took place during that time, there is no single
"Dowland lute" which we can confidently say is the "correct" instrument
with which to play his music. Most of his solo music was composed
before 1600, and it is divided almost equally between six- and
seven-course lutes. At this time use of an octave string on the sixth
course (and possibly higher) was common. Since we have almost no
Dowland autograph scores, authoritative texts are lacking for most of
his solo lute music, and we cannot know how (if at all) he played his
great lute solos in the first two decades of the 17th century, though
we do know that by that time he was playing a nine-course lute, with a
double first string, and ten tied frets on the neck. There is every
reason to believe that by 1610, the actual pitch of a lute "in G" was
somewhat lower than modern pitch, probably about a tone below.
Bibliography
Harwood, Ian, A case of double standards? Instrumental pitch in England c1600. Early Music, 9(4), 470-481. (1981)
Spring, Matthew, The Lute in Britain: a history of the instrument and its music (Oxford, 2001).
Martin Shepherd, 2008