Among the late notes Joyce prepared for typescripts and proofs of Ulysses in 1921 in the V.A. 2 notebook from which he directly copied into his work in progress without any prior rearranging and sorting we find, under the heading “Oxen of the Sun”, a three-word note: “smell moonflower conceive”. This entered the text of Ulysses on a placard (galley proof) between 21 and 29 October 1921 as part of a longer addition describing unusual ways of conception: “by the reek of moonflower or an she lie with a woman which her man has but lain with, effectu secuto or”. The passage is attributed to Stephen Dedalus and the change from “smell” to “reek” is typical of him. He uses the word twice in other sexualised contexts (7.927 or 9.654).
The OED lists several plants called “moon-flowers”, but only “tropical and subtropical climbing plants of the genus Ipomoea” are known for their fragrance. An 1887 quotation reads: “The odorous moon-flower gave out a rich perfume.”
Looking for the suggested superstitious effect on female conception, Joyce’s most likely source turns out to be Havelock Ellis’s Studies in the Psychology of Sex (1905), vol. 5, p. 228, where the moonflower is referred to in a footnote:
Thus the gypsies say of an unmarried woman, who becomes pregnant, “She has smelt the moon-flower”– a flower believed to grow on the so-called moon-mountain and to possess the property of impregnating by its smell.
Ellis’s citation is taken from Hermann Heinrich Ploss and Max Bartels’s Das Weib in der Natur- und Völkerkunde (revised sixth edition of 1899), who in turn quote and rephrase Wlislocki, Aus dem inneren Leben der Zigeuner (1892).
The second part of the addition, homosexual activity leading to pregnancy: “[… ] or an she lie with a woman which her man has but lain with, effectu secuto” (14.245), based on Joyce’s note (NLI notebook 4): “woman impregnated by married woman”, is also derived from Ellis (p. 163):
The stories, repeated in various books, of women who have conceived after homosexual relations with partners who had just left their husbands’ beds are not therefore inherently impossible. [Footnote] The earliest story of the kind with which I am acquainted, that of a widow who was thus impregnated by a married friend, is quoted in Schurig’s Spermatologia (p. 224) from Amatus Lusitanus, Curationum Centuriæ Septem, 1620.
That Joyce was indeed familiar with volume 5 of Ellis’s pioneering work has been rigorously demonstrated by Ronan Crowley in his article ”Looking at Animals without Seeing Them: Havelock Ellis in the ‘Circe’ Episode of Ulysses”, Centre for Manuscript Genetics, University of Antwerp 2017, and Vincent Deane’s recent article “Molly’s taittering lips” discusses another example.
Curiously, Sylvia Beach, Joyce’s publisher for Ulysses, happened to be Havelock Ellis’s Paris agent for Studies in the Psychology of Sex. Ellis sent her a copy of his work in the summer of 1921 to be exchanged for a copy of Ulysses once it was published, as he thought he could not afford Ulysses otherwise. He received copy 295 the year after. Whether Joyce jumped at the chance when the set of six arrived and only found time to extract material from volume 5 or whether he had had access to it earlier is an open question.