By 1603, the famed Elizabethan lutenist John Dowland was so closely associated with his most popular composition, the ‘Lachrimae’ pavan, that he was already signing his name as ‘Jo: dolandi de Lachrimae’. (Figure 1) The very fact that Dowland felt it appropriate to do so in a Continental book indicates that the fame of this work was not just restricted to his home country of England. The widespread dissemination of this piece is unsurprising for two reasons; firstly, Dowland travelled extensively as one of the most sought-after lute virtuosi of his age, holding various posts in Germany and Denmark and, secondly, the vogue for English dance music spread rapidly throughout the German-speaking courts of Northern Europe during the later years of the sixteenth century. Until now, however, no systematic attempt has been made to assess the legion of ‘Lachrimae’ settings and imitations that materialised during the late sixteenth and early seventeenth centuries. The aim of this study, then, is to collate as much of this material as possible and present some preliminary hypotheses regarding in particular the transmission of this piece across mainland Europe. And although a number of ‘Lachrimae’ arrangements elude classification, several ‘families’ of settings nevertheless emerge, including one derived from a distinctive English G minor lute piece (perhaps originating from Dowland himself) which was transmitted in both Continental and English sources. Elsewhere across Europe, a number of interesting lute settings arose which appear to have their roots in lost consort versions, whereas, in other instrumental and vocal repertories, ‘Lachrimae’ became reduced to a two-part framework, often used as a harmonic template for virtuosic elaboration.
The ‘Lachrimae’ pavan is perhaps best known as either the opening pavan of Dowland’s seminal consort collection LoST (1604) or in song form, as ‘Flow my teares’ from 2nd Booke (1600). Robert Spencer (amongst others) has convincingly argued that the latter is a texting of an earlier instrumental model, something which is strongly supported by the fact that the song bears the subheading ‘Lacrime’. Whilst an ‘original’ model for the song cannot be firmly identified (and is quite probably a chimerical notion), a number of earlier settings of the pavan survive in English lute sources. These can be subdivided according to their tonality and warrant a brief summary.
Although it is by no means a certainty, there is circumstantial evidence to suggest that a G minor setting preserved in a number of English manuscript sources may have originally emanated from Dowland himself. The earliest sources for this setting are Dd.2.11, 81v and Dd.5.78, both of which are thought to have been compiled in the late 1580s or early 1590s by Matthew Holmes; later sources include Euing, 31392, JP, Board, Welde, and Herbert. (See Example 1.) On the whole, the piece is transmitted with remarkable stability, with most variants appearing to be ‘performance-orientated’ or minor scribal errors rather than attempts at recomposition. For instance, rhythms are often dotted in later sources (especially throughout bars 11a-14a), cadential formulae slightly varied, and pitch inflections and chord voicings are occasionally the subject of minor alterations. In some later sources, the piece has been adapted for an instrument with additional bass strings. Unusually, the divisions on each strain of the pavan are reproduced with great consistency, the only exceptions being ML (which has some added flourishes) and 6402 (which omits the divisions altogether).
Given the stability of this piece’s transmission, it is interesting to note that Dd.2.11 bears Dowland’s signature (although not alongside any of the ‘Lachrimae’ settings) whilst Board is thought to be connected with Dowland’s teaching activities and contains some autograph material. Furthermore, FD contains a very closely related version of the piece which differs only in that it features an alternative A division and a variant of the final four bars of the C division. Since this source includes five pieces signed by Dowland and is thought to have belonged to a student of his, it seems plausible that this version may be another of his own creation. It is also worth noting that the earliest firmly datable version of this piece, that printed from wood-blocks in Barley (1596), is a G minor setting of a similar ilk to those already discussed. It is not always clear whether the many variants contained in this print arise from typesetting errors or constitute genuine attempts at recomposition. (See Figure 2.) However, Dowland’s famous complaint in the following year that his pieces had been ‘lately printed without my knowledge, falce and unperfect’, whether musically or financially motivated, implicitly suggests that there was at least one ‘approved’ version in circulation which stemmed directly from the composer.
Of course, the proliferation of sources for the G minor version does not necessarily mean that it is the ‘original’; the fact that this version does not fit altogether satisfactorily onto a 6-course lute (g-d-a-F-C-G) might be seen as a further reason to doubt this. The problem of enforced registral displacement of the bass line between bars 13 and 14 of the G minor versions (the low F is not available on a six-course lute) is avoided in A minor settings (Example 2), although this key requires higher hand positions throughout and generally asks more difficult stretches of the player. Certainly, there were A minor versions dating from at least the same time as the early G minor sources, with a unique A minor setting with divisions also occurring in Dd.2.11, 76v (a C minor solo bandora version also appears in the same source (Dd.2.11, 84v)). Two sources, Hirsch and the fragmentary 2764, preserve another, possibly earlier, A minor setting. Whilst this, an ABC setting without divisions, is in some respects simpler than the G minor one, they share certain characteristic features, such as the upper auxiliary in bar 2 (omitted in ‘Flow my teares’ and its derivatives), the inner parts to bars 1 and 4-6, and the major triad at bar 11iii (LoST has a minor triad here). Either (or neither) lute setting might be Dowland’s original, with no convincing reason to prefer one over the other.
As has already been stated, Dowland’s frequent travelling and his fame as a performer resulted in a wide dissemination of his works overseas. Peter Holman has suggested that the 5-part consort version of the ‘Lachrimae’ pavan in Kassel is a pre-publication copy of the LoST version, perhaps circulated by Dowland around 1594-5 when he was in the service of Moritz, Landgrave of Hesse (1572-1632), and there is no reason to suppose that his lute pieces were not similarly distributed. Another route of transmission would be through the lutebooks of travelling noblemen (such as Lord Herbert of Cherbury, whose lutebook happens to contain a late copy of the English G minor version); if foreign pieces were added to such a book whilst the owner was abroad, it is quite possible that some of the English contents would also have been copied by local musicians.
It is interesting to note that there are no surviving Continental sources of the early English A minor lute setting. (In fact, there are no Continental lute versions in A minor at all; even the lute part to LoST, which constitutes a perfectly respectable solo setting in its own right, is not amongst those LoST parts reprinted by Van den Hove in his Delitiae Musice). The G minor version, however, fared slightly better, being both directly copied (although, curiously, without the divisions so popular in England) and used as the basis for further recomposition. The version in Thysius (compiled ?1620s) is a testament to the longevity of this version, being a very close copy of a piece that was by now well over a quarter of a century old. The adjacent page to this includes another lute part apparently in D minor; it is presumably a duet part for a different sized instrument pitched a fifth lower or a fourth higher, unless it is a simple consort part for a D minor setting analogous with those in Morley and Cambridge Consort.
Although not primarily a musical source but a patriotic account of political conflict between the Netherlands and Spain, Valerius includes many songs and instrumental pieces, amongst which are three versions of ‘Lachrimae’ presented together on adjacent pages. These pieces (a two-part vocal version in D minor, a cittern piece in D minor and a lute setting in G minor) were clearly not intended for simultaneous performance, since they all carry the melody (with varying degrees of elaboration and differing harmonic detail). Simon Groot has recently shown that much of the music in this print is taken from printed sources either from England or with strong English connections, and has suggested that the cittern parts were produced either by Valerius or by someone within his circle, since they are derived from the vocal versions of the melodies. This may well be the case in some instances, but the ‘Lachrimae’ setting contains a bar rest during the imitative passage in strain C (bar 20), which strongly suggests that this unusual piece belongs to an ensemble arrangement of some description. There can be little doubt that the lute version is a second- (perhaps third-) generation derivative of the English G minor setting, being replete with printing errors (bar 3 opens with an incorrect chord) and recompositions (the flourish from bar 2iii, or the ornamented sequential passage from bar 12), yet clearly a close relative.
Romers also contains a G minor ‘Lachrimae’ which has much in common with the English sources, particularly the version in 6402. Whilst the melodic contour of the English piece has been both eroded (bar 4) and elaborated (bar 6) in places (Example 3), the Romers version is largely harmonically consistent with its probable model, and displays the characteristically English auxiliary note in bar 2. There are some odd substitute chords in bar 9 and bar 22-23 but these, in all likelihood, result from a scribe attempting to make corrections when copying from an exemplar in which some of the bass notes have been positioned on the wrong strings, rather than representing a genuine harmonic innovation (Example 4). Although little is known for sure about this manuscript, it would appear to be of Westphalian origin, and the numerous English pieces might well have been imported via Kassel, since it includes works by ‘M[oritz] L[andgraf von] H[esse]’.
Another instance of a Continental lute arrangement derived from the English G minor version can be found in Fuhrmann, ascribed to V[alentin] S[trobel]. Although featuring an elaborate and unique set of divisions, this setting’s model is quite apparent. Most notably, this would appear to be an example of a Continental version being exported back to England, since, as well as the south-German copy in Haslemere, it appears (alongside a good deal of other Continental music) in Cosens, 36v, which is certainly of English provenance. In this source, however, it is initialled ‘C.K.’
An interesting approach to the English G minor setting can be found in the Thesaurus Harmonicus of the French lutenist Jean-Baptiste Besard (Cologne, 1603), who spent much of his career in Germany as a lute teacher. Despite its title, his ‘Fantasia Ioannis Doland Angli Lachrima’ (Figure 3) is by-and-large a version of the English piece with ‘improvements’ added to its divisions. Besard’s choice of title possibly provides a degree of insight into his view of the ‘Lachrimae’ pavan as a starting point for compositional and performative elaboration rather than as a sacrosanct entity. The most notable melodic deviations from its English model can be seen in the A strain, whilst a number of interesting harmonic substitutions are evident throughout. The most notable of these are a) the strong bass progression in bar 2 (something that Besard later employed in the odd consort version included in his Novus Partus) and b) the 6/4 chord in bar 20iii (Example 5). However, this piece, like much of Besard’s printed output, is also riddled with misprints (especially rhythmic ones), which often render it difficult to discern between error and invention. To single out just one instance, the highest pitch of the very first chord is carelessly omitted, thus ruining the famous descending tetrachord. (See Figure 3.) Bearing in mind the early date of this setting and the popularity and influence enjoyed by Besard, it is easy to imagine how the ‘Lachrimae’ pavan metamorphosed so rapidly and radically. This process would have been exacerbated by the copying of flawed printed versions into manuscript anthologies with all errors intact; the direct copy of this one in Nauclerus serves as an excellent case in point.
Besides the derivatives of the English G minor version that were in circulation on the Continent, there were a number of interesting lute settings with no apparent connection with surviving English sources. The earliest firmly datable Continental setting is the ‘Pavana a 5 voc. Dulandi Angli’ published by Rude in 1600. (Example 6) The somewhat intriguing rubric attached to this piece seems to suggest that it is an intabulation of a lost consort setting, something which is further supported by the fact that a point of melodic imitation disappears from the texture during bars 20-21. However, one should take care not to assume that this is necessarily a successful intabulation of its model, nor that it is an intabulation of a good arrangement. Indeed, the clumsy attempt at an inner voice suspension in bar 4iii-iv (perhaps a misprint?), not to mention the uniquely condensed / adjusted harmonic rhythm of bars 21, suggest that, somewhere earlier along the evolutionary line, a less-than-satisfactory source may have contributed to this particular version.
A few other lute settings (Dlugoraj 78, Königsberg 24, Brussels), all of which are in G minor, lack divisions and bear little in common with their English counterparts, can be fruitfully considered as a group alongside Rude’s version. This is a disparate group, with no clear sequential relationship discernible between them (although this is perhaps unsurprising when one considers that their period of compilation spans approximately 35 years). Nevertheless, a number of salient features are shared between these versions and Rude, none of which are encountered in the English transmission(s).
To give only a handful of examples, all four feature a substitute chord (an inserted IV) in bar 2 unseen in any English sources (Example 7a). Furthermore, Rude’s distinctive ascending bass motif in bar 19 is also featured in Königsberg 24 (Example 7b), whilst the exposed descending run in bar 20 is mirrored in both Königsberg 24v and Brussels. All four versions seem to share from a common fount of elaborative melodic material, often resulting in a marked erosion of the original ‘Lachrimae’ melody; particularly characteristic are the prominent melodic auxiliary notes in bars 1 and 3 (Brussels, Dlugoraj 78) and bar 9 (Rude, Königsberg 24 and Dlugoraj 78) (Example 7c & Example 7d). Whilst these four sources do not by any means share all of the same features, the web of cross-references between them strongly suggests that this group might have either derived from Rude’s intabulation, or be based upon the same distant original model.
It is worth mentioning briefly the two versions in Montbuysson 5 and Montbuysson 55v, both of which display a closer stylistic affinity with the G minor English setting than the previously discussed sources, whilst simultaneously exhibiting some of the features which we have identified as being distinctly ‘Continental’ (e.g. the important chord substitution in bar 2). Montbuysson was based at Kassel from 1598 to 1627, where he might well have encountered (either in manuscript or performance) English versions of the piece stemming from Dowland himself (who had been employed there in 1594). The two closely-related settings in this source were not committed to paper until 1611, and thus might well represent a composite of a number of English and Continental versions known to Montbuysson. The melodic simplifications incorporated into these pieces and their relative simplicity might well have arisen from the pedagogical function of Montbuysson’s manuscript.
Two more interesting Continental lute settings lie in close chronological proximity to Rude’s, namely the one which is included in Joachim van den Hove’s Florida (1601) and the anonymous setting in Herold, according to its title-page a copy of a German lutebook made in 1602 by a student studying in Padua. There are some exterior indications that there may be a connection between these two versions, since both are in G minor, are coupled with triple-time versions, and use the unusual formal scheme AAiBCBiCi. Most notably, perhaps, both are exceedingly florid throughout, with little attention paid to the melodic contour of the English pavan. Furthermore, close examination of the undivided strains reveals a greater similarity between the two versions than initially meets the eye or ear, although the divisions are largely unique. The undivided strains employ essentially the same structural pitches throughout, suggesting that they are derived from a common harmonic template. Both settings also exhibit a penchant for virtuosic writing in parallel thirds and sixths and, on occasions (such as between bars 3-8) even duplicate passagework more-or-less verbatim (Example 8a). Another interesting feature almost unique to these two versions (only the version for violin and bass instrument in Schermar has it too) is the dominant chord substituted into bar 10 (Example 8b). These musical details suggest very strongly that the Herold setting is also the work of Van den Hove, something which is further supported by the fact that many of the pieces located nearby in Herold are either ascribed to him or were included in Florida a year before Herold was copied from its exemplar. Perhaps, then, these settings might be thought of as two contrasting yet more-or-less contemporary ‘snapshots’ of one performer’s use of the pavan as a framework within which to display his virtuosity. Even at this early stage, the melody of the English version is disappearing and the piece is beginning to be treated as a chord sequence , serving as a basis for further elaboration. However, by the time Van den Hove added another setting to the Schele lutebook (dated 16th February 1614), he had obviously experimented with other models, since the unorthodox AAiBCBiCi layout has been replaced with a more conventional AAiBBiCCi and the contrapuntal template differs somewhat. Virtuoso passage-work remains prominent throughout however; this setting is one of the most elaborate of all lute versions.
After Barley’s lute arrangement, the next earliest firmly datable ‘Lachrimae’ setting is the mixed consort arrangement published in Morley (1599). This collection obviously acheived considerable popularity as it was reprinted in 1611, whilst some of the bandora parts (including that for ‘Lachrimae’) were copied into Königsberg 38v. Morley’s version appears to be closely related to that found amongst the Cambridge Consort manuscripts, the excellently-crafted lute part of which (Dd.3.18) has been used by Sydney Beck in lieu of the missing Morley lute partbook. This part also seems to have been the basis for the curious ‘contra’ part, not, presumably, to be played on the lute, of an incomplete consort-song version that was appended to Thomas Wode’s partbooks some time after 1610. (Example 9)
It was as a song, however, that the ‘Lachrimae’ pavan really seems to have become the ubiquitous ‘hit’ of its age. ‘Flow my teares’ was published in 2nd Booke (1600) as two texted vocal parts with a lute accompaniment, although it enjoyed a lengthy life as a solo continuo song without the lute part (Tenbury, Earle, Och 439 (actually a variant version in D minor), Forbes (3 eds. from 1666-1680s). Indeed, the transmission of the ‘Lachrimae’ pavan throughout the seventeenth century can be understood as stemming almost exclusively from the publication of this song. There are perhaps two main reasons for this, the most obvious of which is the size of its print-run; one thousand copies was immense for this period (and the publishers presumably expected to sell every copy). Secondly, the scoring of the song arrangement transformed the piece from a somewhat tricky lute piece (or something that required an instrumental ensemble to perform it) into a contrapuntally coherent two-voice entity, despite the written-out lute part. Thus, not only was the ‘Lachrimae’ pavan more readily available as a text, but it was ironically now more accessible to performers without a lute at hand. As we shall see, the simple two-part reduction offered great potential to both composers and performers.
An early example of the exploitation of this model stems from none other than Dowland himself. The ‘Lachrimae Antiquae’ of LoST (and its related versions in Kassel and Melville) is clearly based upon the two vocal parts from ‘Flow’, the inner parts presumably worked into place afterwards. Several clues suggest this dependancy upon the song, not least a handful of melodic details which mirror the syllabic patterns of the texted cantus part (e.g. the reiteration of the melody note of bar 3iv (‘ev-er’) and the exquisite setting of ‘sad in-fam-y’ in bar 6) (Example 10). The previously-discussed auxiliary note in bar 2, so typical of the English lute versions but a notable absentee in ‘Flow’, is also omitted here and, in most instances, the registral shifts and use of accidentals in LoST match those of ‘Flow’. Craig Monson has convincingly argued that a similar creative process resulted in ‘Mr Dowland’s Lachrimae’, a D minor consort setting attributed to William Wigthorpe, as well as a consort-song version of ‘Sorrow stay’ in the same source; both appear to be based upon the respective vocal parts printed in 2nd Booke with inner parts added later.
The creation of new settings from the voice parts to ‘Flow’ has been traced by Werner Breig in a study of various Northern European keyboard settings (Byrd FVB, Sweelinck and Schildt; a few more are discernible). But besides being employed as a compositional framework for elaborations of this nature, it is important to note that a number of other more modest 2-part versions, both instrumental and vocal, were circulated across mainland Europe. Valerius, for instance, included a double-texted 2-part version in D minor. Although the cantus in particular is highly ornamented, it bears a close resemblance to the Wigthorpe outer parts in the same key which, of course, would appear to stem from 2nd Booke. Valerius could have based his parts directly on 2nd Booke, or one of any number of derivatives that may have been circulating in manuscript. Camphuysen 1652 gives another 2-part version, this time with a devotional Dutch text, supplied with instrumental divisions (for both cantus and bassus) by one Joseph Butler, a Londoner working in Amsterdam. Clearly then, whilst the melody and bass of the ‘Lachrimae’ pavan (and, implicitly, the harmonic content of the piece) continued to be transmitted, the common fount of inner parts employed in the earliest English lute and consort settings had ceased to be considered of much importance.
A number of interesting European instrumental versions also survive from the mid-seventeenth century. Many of these suggest that ‘Lachrimae’ was no longer considered to be a typical pavan and was now considered to be almost an instrumental genre of its own. The unusual ABC, AiBiCi (etc.) scheme of Van Eyck 12 (1648) certainly seems to suggest that instrumental divisions were now considered more important than the formal scheme of the original pavan. Furthermore, in the same year that van Eyck’s collection was published, its dedicatee, the polymath and expert musician Constantijn Huygens, was enjoying the company of the virtuoso viol player Ditrich Stoeffken in The Hague and wrote of his delight in accompanying him in terms that suggest the piece was used for extempore divisions:
‘… Mr Stöfkens and I are doing a kind of wonders upon two Leereway Viols … in time of necessitie I have hands enough to play a wofull Lachrime and such other stuffe upon my organs to have that wonderfull bowe rowle upon my Bases.’
Elsewhere, two-part arrangements designed to show off the newest developments in violin technique had begun to appear, such as the modest one in Schermar (1620) and Schop’s contribution to t’Uitnement Kabinet (1646). The latter is particularly interesting since, beyond the opening tetrachord descent, its ‘melody’ bears little resemblance to that of either the early lute version or either of Dowland’s published versions. A similar situation can be observed in some later lute settings, such as Stobaeus (possibly as late as the 1630s?); the melody has become eroded to the point that ‘Lachrimae’ is largely just a suggestive chord sequence, albeit one with certain motivic and contrapuntal conventions attached to it. (See Example 11.)
Nevertheless, ‘Lachrimae’ continued to be considered fertile ground for variation-writing long into the seventeenth century. A keyboard setting by Schildt survives in three sources, one of which (Clausholm - possibly autograph?) is dated 1634 and another can be dated from after 1642 (Voigtländer); the fine version by his teacher Sweelinck, survives in only one source, a rather corrupt tablature possibly dating from as late as the 1670s. These settings, along with that attributed to another Sweelinck pupils, Scheidemann, are particularly flamboyant, each featuring complex (and unique) divisions. In Schildt and Scheidemann, the melody is liberally decorated even in undivided strains, with the floridity of Schildt’s treatment being comparable with that of Van den Hove’s. Indeed, in many respects, these keyboard settings are the aesthetic bedfellows of Van den Hove’s lute arrangements, indulging in similar passage-writing in parallel thirds and sixths and displaying the same penchant for instrumental showmanship throughout. However, the keyboard composers are able to exploit a contrapuntal dimension unavailable to Van den Hove (and lutenist-composers in general), allowing them to make imitative interplay of paramount importance in their settings. Schildt and Scheidemann also experimented with sesquialtera passages, something Van den Hove himself dabbled with briefly in the final division of Schele, and which is briefly foreshadowed towards the end of Dd.2.11, 76v. Schildt’s and Scheidemann’s settings, then, not only share the virtuoso aesthetic of van den Hove’s lute solos but give a strong impression of being compositional tours de force as well.
The ‘Lachrimae’ pavan was evidently a bona fide ‘hit’ of its age. Unfortunately, such are the huge number of settings, derivatives and imitations it spawned that only a small sample can be fruitfully discussed here. Furthermore, we are undoubtedly left with an incomplete picture of what was once in circulation. We know, for instance, of several missing consort versions from single parts preserved in Dlugoraj 122, TCD Tallis, the Browne bandora book and the Valerius cittern part, as well as from the title to Rude’s lute setting. Several perplexing unica serve to emphasise the incompleteness of the surviving picture, not least the fascinating F minor lute setting in Eysertt, which, although tangential to the patterns of transmission identified in this paper, nevertheless seems to be a good, error-free version. Dowland, of course, was in Nuremburg in 1595, so it is not beyond the realms of possibility that this piece may have originated from him in some sense. Of course, not all sole survivors are so musically satisfying; the consort version which might have accommodated Valerius’ bizarre cittern part is perhaps best left to the imagination, while the odd 4-part vocal setting transmitted in Camphuysen 1647 and Camphuysen 1655 and Besard’s Novus Partus arrangement (devoid of any trace of the melody, or any other melody, for that matter …) remain inexplicable oddities for the time being.
This study has only scratched the surface of a large topic and would be hugely enhanced by similar research into the multitude of similarly-transmitted English pieces that were popular across late-sixteenth and early-seventeenth century Europe. By cross-referencing our hypotheses regarding the dissemination of ‘Lachrimae’ settings with the results of similar studies, a much fuller picture of cross-repertorial interchange could be achieved, shedding a good deal of new light over patterns of musical import- and exportation and perhaps providing new insights into the compilation of some of these sources.
Although this research was carried out within the context of a computer-assisted electronic corpus-building project, the analysis of the musical material has been carried out entirely manually. Electronic resources have been used, certainly, but mainly for auditioning and printing out the 60 or so versions of ‘Lachrimae’ discussed in this paper. Our corpus currently contains over 90 settings of Dowland’s pavan and closely-related pieces. The kind of detailed comparative examination we have begun here on a modest subset of this corpus would only be feasible on the full collection (let alone larger corpora) with the aid of computer analysis software tools. A principal aim of this work has been to identify the types of comparison and analysis that might in future be done by machine (and to define the limits beyond which machine-analysis can, or should, not venture), and to provide some kind of benchmark against which such development might be assessed. It is our belief that the work reported in this paper represents about the limit that could be undertaken manually within a reasonable period of time. For larger-scale repertory studies, such as a long-overdue assessment of all English instrumental music in 17th-century Continental sources, the laborious work of bar-by-bar comparison could be much eased by intervention from analysis software. This means that the intense human labour could be devoted to the major task of data-entry and subsequently to the subjective interpretation of the results of the analysis.