Date and circumstances of composition. After the aborted attempt at publication of the 1896 version of The Shadowy Waters (edited in vol. 2 of present edition), WBY's work on this long-running project seems to have faltered for some time, before being taken up again in earnest in May 1899. The verse drama was nevertheless much in the poet's thoughts in the interim, and his struggles with the work were announced to AG as early as 1897, as she noted after a tea with WBY in London (entry for 23 Feb. 1897, AGD92–02, 129):

He [WBY] believes there will be a reaction after the realism of Ibsen, and romance will have its turn – He has put “a great deal of himself ” into his own new play – “the Shelter of the Waters”? – and rather startled me by saying about half his characters have eagles’ faces –

In the course of 1897, WBY continued to think about revision and new composition in the piece, which by the summer he was still considering as his contribution to a newly founded ‘Celtic Literary Theatre’ involving both AG and Edward Martyn. In a letter of 7 Jul. 1897, JBY told his friend Sarah Purser about his son's play, a version of which he had by now evidently read (quoted CL 2, 121):

Next summer some sort of an association with some sort of a Celtic appellation is going to give in Dublin a series of theatrical entertainments under the management of Mrs. Emery – they are to present a play by Edward Martyn (of Tillyra Castle) and Willie's play called the “Shadowy Waters” this latter to me absolutely unintelligible – however Mrs. Emery [Florence Farr] says she understands it and as she is to act in it one of the principal parts this is important though I suppose not absolutely necessary.

Even in its 1897 form the play was evidently quite unsuitable for the stage, and in the event WBY's The Countess Cathleen fitted the new group's theatrical purposes much more readily. In Jun. and Jul. 1897, while based at Tillyra as Martyn's guest, WBY is likely to have discussed his poetic play in some detail with GM, whose later recollection (which may also draw on discussions the following summer) gives some flavour of the conversations (Ave (1911), 241):

[WBY] had come over to Tillyra from Coole a few days before, and had read us The Shadowy Waters, a poem that he had been working on for more than seven years, using it as a receptacle or storehouse for all the fancies that had crossed his mind during that time, and these were so numerous that the pirate-ship ranging the Shadowy Waters came to us laden to the gunnel with Fomorians, beaked and unbeaked, spirits of Good and Evil of various repute, and, so far as we could understand the poem, these accompanied a metaphysical pirate of ancient Ireland cruising in the unknown waters of the North Sea in search of some ultimate kingdom. We admitted to Yeats, Edward [Martyn] and I, that no audience would be able to discover the story of the play, and we confessed ourselves among the baffled that would sit bewildered and go out raging against the poet. Our criticism did not appear to surprise Yeats; he seemed to realize that he had knotted and entangled his skein till no remedy short of breaking some of the threads would avail, and he eagerly accepted my proposal to go over to Coole to talk out the poem with him, and to redeem it, if possible, from the Fomorians. He would regret their picturesque appearance; but could I get rid of them, without losing the poetical passages? He would not like the words ‘poetical passages’ – I should have written ‘beautiful verses’.

WBY established himself at Coole on 26 Jul. for a stay of two months; but there is no direct evidence on further composition of the play at this time. GM was, however, to have further involvement with the composition process, and WBY was to read him the play aloud (in the company of Arthur Symons, JBY, and Edmund Gosse) on two occasions at Christmas, 1898. Whatever the outcome of conversations at this time, sustained work on rewriting had to wait for WBY's stay at Coole in May 1899. Now, as the editors of DC comment, ‘The conditions [for composition] were ideal; Lady Gregory [. . .] would protect his health and his time,’ while ‘His theatre colleagues, Moore and Martyn, were near at hand, the former ready to give the practical advice that proved invaluable in the inauguration of the Irish Literary Theatre’ (DC, 225). There seems little doubt that WBY was spurred on at this point by thoughts of TSW as a potential stage-play, which would also constitute a poetic work for publication. The recent success (and accompanying public controversy) of The Countess Cathleen in Dublin gave WBY reason to attend to news that in London there were plans to establish the Stage Society, which would ‘serve as an Experimental Theatre’. Writing to Clement Shorter from Coole on 27 May 1899, WBY announced his work to ‘finish’ the play (CL 2, 418):

As a result of the success of ‘The Irish Literary Theatre’ I have a chance of getting ‘The Shadowy Waters’ done in London in Autumn and am therefore setting to work to finish it. I told you about it I think. It is a rather wild little play about the length of ‘The Land of Heart's Desire’, which acted rather less than 25 minutes, and probably the best verse I have written. Do you think a play of this length is too long for a magazine, American or English? I should want to get it published before it was acted if possible.

It would appear from this that WBY had relatively quick completion in mind, and this may be owing in part to detailed conversations about the play which had taken place with various friends including AG and GM over the past couple of years. Word of the play was passed to the Daily Express in Dublin, which reported on 10 Jun. that it was to be produced as a curtain-raiser in London that autumn, and that ‘Mr. Yeats thinks it the best thing he has written’ (quoted in CL 2, 438). The pace of work was maintained over the coming weeks: on 21 Jun. WBY reported that the play ‘is going on far better than when I left it aside a couple of years ago’ (to Dora Sigerson Shorter, CL 2, 425), and on 12 Jul. the poet wrote to SMY about how “The Shadowy Waters’ is not finished, but is going on well’ (CL 2, 433). As the summer went on, further reports confirmed both WBY's dedication to the work in hand (‘I am deep in a long poem, which I dare not interrupt’, he wrote to John Lane on 3 Aug. (CL 2 435)), and his confidence in its quality (writing to JBY on 11 Aug., ‘it [TSW] is going very well [. . .] in some ways the best long poem I have done [. . .] more intense and more original’ (CL 2, 437)). Plans for publication began to take shape in Aug., and WBY's sights were set on the North American Review (NAR), to whom he had by now (and lucratively) sold an essay, ‘The Literary Movement in Ireland’ (publ. Dec. 1899; CW 9, 459–470). The magazine had a London office, and the editor, William D. Fitts, who had already asked the poet whether he might be able to provide some verse, was in England that summer. WBY's letter to Fitts of 19 Aug., offering him TSW, anticipated completion of the piece in Sept. or Oct. (CW 2, 439–440):

I am at work on a dramatic poem, ‘The Shadowy Waters’, which has to be done by September, if I am to get it on the stage in London, which I believe I can do this autumn. I cannot stop from this poem long enough to do you any poem of sufficient importance for me to send you for the generous terms you offer. I am a very slow writer. I have never done more than five or six good lines in a day, and not often that, and a poem of thirty or forty would take a couple of weeks probably. I suggest in my letter to New York that ‘The Shadowy Waters’ may not be too long for ‘The North American Review’. It will be about 700 lines when finished. I have written about 400 lines. I don't think it will be a popular poem, but I think it may be a good deal noticed for it is very wild and passionate. It is lyrical in feeling, and will contain a certain number of actual lyrics. It is what people call Maeterlinckean, though certainly it owes nothing to him. The subject is old Irish, and it is of course in one act. I am anxious to impress on you that I do not think it will have many popular qualities, but I think it is my best poem of any length. It is quite unlike anything I have done. If you cared for it you could have it some time in October. If it is acted it will I think be so in November.

Since the poem as eventually published is 431 lines long, it appears that the bulk of actual composition (though not, of course, detailed revision) had in fact been done by this point. What the projected 300 extra lines might have contained is not clear, and it seems unlikely that these would have been made up entirely of lyrics (in the event, no lyrics at all featured in the play). When WBY informed George Russell about the work at the end of Aug., he approached the subject by way of symbolism, of Russell's vision of the ‘white fool’, and the god Aengus: ‘I may be getting the whole story,’ WBY wrote, ‘of the relations of man and woman, in symbol, – all that makes the subject of ‘The Shadowy Waters" (CL 2, 443). Yet even the appearance of Aengus would prove insufficient to allay pressing anxieties about the literary interference of GM. Russell's enthusiasm for the whole project of revision was very limited; and in Sep. 1899, on learning of some of GM's proposed amendments to the plot, he told WBY how ‘I swore at Moore when I heard it [. . .] I would like to strangle him’. GM's intervention, it should be said, was made with WBY's full consent, certainly during the summer of 1899. Ridding the play finally of its hawk-headed Fomorians was both logical and necessary in order to create a potential script for the stage; and it seems likely that WBY wanted to make this radical exclusion more than just his own decision. One of GM's retrospective glimpses of the re-composition discussions at Coole identifies in WBY a willingness to turn away from ‘human sympathies’ (as instanced by his forthright rejection of new dramatic work by Martyn), and links this with the new shape for TSW (GM, Ave (1911), 282–283):

Yesterevening, when we wandered about the lake, talking of The Shadowy Waters, trying to free it from the occult sciences that had grown about it, Fomorians beaked and unbeaked, and magic harps and Druid spells, I did not perceive that the difficulties into which the story had wandered could be attributed to a lack of human sympathy. But Yeats's treatment of Edward [Martyn] proved it to me. [. . .] To write a play our human and artistic sympathies must be very evenly balanced, and I remembered that amongst my suggestions for the reconstruction of The Shadowy Waters, the one that Yeats refused most resolutely was that the woman should refuse to accompany the metaphysical pirate to the ultimate North, but return somewhat diffidently, ashamed of herself, to the sailors who were drinking yellow ale. ‘Yeats has reflected himself in the pirate,’ I said. ‘All he cares for is a piece of literature.’

Much of the 1899 work on TSW took place at Coole: there was an interruption in Sep., when WBY tailed MG in Belfast and in Dublin, but by Oct. he was back in the care of AG, and busily perfecting the play. The arrival of a playscript from Fiona Macleod, The Immortal Hour, revealed that William Sharp was already taking material and ideas wholesale from WBY, including elements of TSW: presumably, this helped stiffen general resolve at Coole. Also in Oct., Russell had been appraised of the ongoing work, and continued to feel instinctively hostile to it, not least on account of the involvement of GM. Russell told AG that he was ‘sorry to hear that Moore has brought his inartistic soul to bear on the Shadowy Waters and that Willie is again altering’; his intention was to visit Coole and intervene in person: ‘I will strangle Moore if necessary’ (12 Aug. 1899, Berg collection, NYPL). GM and Russell did meet in Dublin at this time, though any entreaties (up to and including attempted strangulation) were in vain, and the collaboration in Co. Galway continued. Replying in early Nov. to a letter from Russell calling GM ‘the fiend who has suggested alterations’, WBY defended the changes that were now being effected (CL 2, 463–464):

I think you are wrong about ‘The Shadowy Waters’. The picture was more impressive in its old form and I regret the loss of the Fomor but the poetry is richer and more various in the new, and it is getting written more easily. The new form will act much better. Moore does not much like my ideas of the proper way of speaking verse; but he is wrong and I want to do a little play which can be acted and half chanted and so help the return of bigger poetical plays to the stage. This is really a magical revolution for the magical word is the chanted word. The new ‘Shadowy Waters’ could be acted on two big tables in a drawing room; not that this will please you who don't much like acting at all I think.

As the year came to a close, WBY was back in London, and racing against a deadline for the revised work. On 21 Dec. the poet told AG how he had ‘been trying to get my poem finished for next Saturday and will almost succeed’, since ‘I am about twenty or thirty lines from the end’: he observed that ‘The thing grows wilder and finer as it goes on, I think’, and his confidence increased further by the next day, as he announced that ‘I have just got to the end of ‘The Shadowy Waters’ and two days’ revision here and there will have it ready for the ‘North American Review’ ’ (CL 2, 479). However, on 19 Jan. 1900, Russell was urging WBY to ‘Please publish ‘The Shadowy Waters’ at once, before any more changes are suggested by the changing Tatuas’ (Denson, 35). The real reason for delay at this point was not in fact anything to do with Tattwas Tarot card symbolism, but the death of WBY's mother on 3 Jan., which set back the poet's practical plans for having a finished TS ready for NAR. Nevertheless, WBY was able to contact G.G. Leveson-Gower (the NAR's European editor) on 8 Jan., apologizing for not having given him the poem on 5 Jan. as agreed, and explaining this by ‘an unexpected family trouble’, and letting him know that the typed copy would be delivered on 9 Jan. A copy of the TS was with the New York editor of NAR, William Fitts, by the middle of Mar., and TSW was published in the magazine's May edition.