Charles Ricketts’s illustration for ‘The Selfish Giant’ in the first edition of Oscar Wilde’s The Happy Prince and Other Tales (1888)
Oscar Wilde’s fairy tale ‘The Selfish Giant’ tells of a giant who builds a wall around his garden to keep out trespassing children. A perpetual winter reigns in the garden, until the children enter through a hole in the wall and all the trees break out into blossom. The giant resolves to knock down the wall and allow the children to play in his garden. One of the children reveals himself to be Christ and grants the giant a place in paradise.
The story, first published in the 1888 collection The Happy Prince and Other Tales, is one of Wilde’s most popular. But did Wilde really write it?
This question was first posed by Wilde scholar – and Wilde’s only grandson – Merlin Holland. In 2008 Lúcia Moreira Salles, a Brazilian collector, donated a set of Wilde manuscripts to New York’s Morgan Library & Museum. Holland, who had been invited to write a brief book on the manuscripts, noticed that one of the items, a complete fair copy manuscript of ‘The Selfish Giant’, is in Constance’s rather than Oscar’s hand. Oscar had made a few pencilled edits to the manuscript and added a title page identifying the story as ‘by Oscar Wilde’.The manuscript can be viewed here. Holland, M. (2008) A Portrait of Oscar Wilde, Privately printed.
Holland wondered what this manuscript signified. He thought it unlikely that Oscar had dictated the story to Constance, as this would have impeded the flow of Oscar’s storytelling. Perhaps Constance had composed the story herself. Holland considered this idea plausible given that Oscar was not the only member of the family to publish a collection of fairy tales in 1888. That year also saw the publication of Constance’s volume of children’s stories, There was Once. Constance’s letters to her brother dating to this period are full of references to stories she had written or intended to write.
For Holland, ‘The Selfish Giant’ is quite different to the remaining stories in The Happy Prince and Other Tales. It lacks the witty conversation that is a hallmark of Wilde’s writing and that can be found in abundance in ‘The Happy Prince’, ‘The Remarkable Rocket’, ‘The Devoted Friend’, and ‘The Nightingale and the Rose’.
Holland noted that there are substantial differences between Constance’s manuscript and the version that appeared in The Happy Prince and Other Tales. For example, when the Christ-child points to his stigmata, in the published version he says, ‘these are the wounds of love’. In Constance’s version he says, ‘this was done many years ago, that all men might be saved’.
Speaking on the RTÉ podcast Wilde Stories in 2016, Holland said, ‘It’s just not the sort of thing that Oscar would have written, that bit in her version. I hesitate to say it these days, because people […] who love the story would say, “the idea that this story, ‘The Selfish Giant’, my favourite, could possibly have been written by his wife, no, absolutely impossible, how dare you say such a thing?” But it’s very much to Constance’s credit, that I think she must have had a little bit of a hand in it somewhere. There’s no doubt about it in my mind.’Wilde Stories 5: The Selfish Giant, https://www.rte.ie/radio/podcasts/21840022-wilde-stories-5-the-selfish-giant-the-lyric-fea
There was Once (1888) by Constance Wilde
Other Wilde scholars have taken up Holland’s theories. Franny Moyle, in her 2011 biography of Constance, notes that the overtly Christian ending of ‘The Selfish Giant’ feels at odds with the more secular tone of the four remaining stories in Wilde’s collection.Moyle, F. (2011) Constance: The Tragic and Scandalous Life of Mrs Oscar Wilde, John Murray, 136–7 Constance was more conventionally religious than her husband and differed from him in her belief that art should have a moral purpose. Ian Small, in his 2017 edition of Wilde’s short fiction, ponders whether the story has its origins in ‘some kind of informal collaboration’ between the couple.Small, I. (ed.) (2017) The Complete Works of Oscar Wilde, Vol. 8: The Short Fiction, Oxford University Press, xxii–xxv
But in the absence of any smoking-gun evidence – letters, memoirs – that would prove one way or the other who wrote the story, the authorship has remained a mystery.
Now a team of scholars from the University of Bucharest has attacked the question from a different angle. In early July 2026 Liviu Dinu, Alina Iacob, and Cosmin Ciotloş presented the results of a stylometric investigation of ‘The Selfish Giant’ at the 6th International Conference on Natural Language Processing for the Digital Humanities, in San Diego.Dinu, L. P., Iacob, A., & Ciotloş, C. (2026) A computational authorship attribution in the children’s tales of Oscar and Constance Wilde: the case of ‘The Selfish Giant’, Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities, 381–9, https://aclanthology.org/2026.nlp4dh-1.34.pdf
Computational stylometrics is a type of research method that involves the mathematical analysis of aspects of a text’s style. It typically relies on features that are difficult or impossible for a human reader to notice simply by reading a text, such as the frequency with which certain pairs of letters appear together. The scholars fed into their model the five stories in Constance’s There was Once and the five stories in Oscar’s The Happy Prince and Other Tales. They found that ‘The Selfish Giant’ is stylistically more similar to Constance’s than to Oscar’s stories.
Figure from Dinu, Iacob, & Ciotloş (2026) showing how ‘The Selfish Giant’ clusters with Constance’s rather than Oscar’s stories.
The scholars conclude that ‘there is a clear separation between the stylometric profiles of the two authors, with “The Selfish Giant” emerging as the only outlier. The strong stylistic resemblance between this story and Constance Wilde’s writings suggests that it may have been the product of a collaboration, or even authored by her alone.’
This seems to be clear and objective evidence for Constance’s authorship of the story. But I am not quite ready to scribble out Oscar’s name on the title page of my copy of The Happy Prince and Other Tales and replace it with Constance’s. Stylometric models are more accurate when they are based on a larger corpus, and the researchers behind this study did not include all of Constance’s and Oscar’s fairy tales. Would the results of the analysis change if more stories were added into the mix?
Also, previous research has suggested that authorship attribution studies based on short texts are unreliable: stylometrist Maciej Eder made the case in 2015 that the most commonly used methods of author attribution perform poorly when the sample length is below 5,000 words and are equivalent to guessing when samples are below 1,000 words.Eder, M. (2015) Does size matter? Authorship attribution, small samples, big problem, Digital Scholarship in the Humanities, 30, 167–82 ‘The Selfish Giant’ is about 1,600 words. I asked the scholars about this point, and Dinu told me that the metric they used – rank distance – takes into account writers’ preference for certain functional words and is not impacted by the frequency with which those words are used: this method is appropriate for texts of shorter length.Liviu Dinu to Rob Marland, personal communication, 22 Jul. 2026
Even if we accept the findings from Bucharest, this does not imply that the story is entirely Constance’s work. Holland observes that Wilde’s poem ‘Le Jardin de Tuileries’ shares an image with ‘The Selfish Giant’. It ends: ‘Ah! cruel tree! if I were you, / And Children climbed me, for their sake / Though it be winter I would break / Into spring blossoms white and blue!’ This poem may have been written when Oscar and Constance were honeymooning in Paris in 1884 and had rooms overlooking the Tuileries gardens. ‘Was the plot [of ‘The Selfish Giant’] outlined to Constance by Oscar, for her to write and him to polish?’ asks Holland.
This theory would be consistent with the stylometric evidence and the fact that most of Constance’s fairy tales are retellings of classic stories, such as ‘Little Red Riding Hood’ and ‘Cinderella’, which suggests that she preferred to work from plots devised by others. Wilde is known to have freely given plots for plays and stories to friends.See, e.g., this letter.
Another possibility, and one that Holland first proposed, is that Constance wrote the story based on one she had heard her husband tell. Several members of Wilde’s circle committed to paper stories that Wilde had told them. After Wilde died his friend Aimeé Lowther published four such stories, including two Christian parables. The memoirs of Richard Le Gallienne, André Gide, and Arthur Conan Doyle also feature retellings of Wilde’s oral tales.
This is the theory I think is most likely to be true, given that the story contains several images that seem more characteristic of Wilde than his wife. For example, the eponymous character says, ‘I have many beautiful flowers, […] but the children are the most beautiful flowers of all.’ Wilde repeatedly compared children with flowers,Wilde told Mary Blakeney, the President of the Women’s Literary Club of Sacramento, that her ten-year-old daughter May was ‘like a wonderful little flower’; to an interviewer in the New York resort of Richfield Springs he said, ‘I love children. They are living flowers. I think of adopting one or two.’ (Holland, M. & Hart-Davis, R. (eds) (2000) The Complete Letters of Oscar Wilde, Fourth Estate, 160; ‘Sizing Us Up’, The Sunday Herald (Boston, MA), 20 Aug. 1882, 12, in Marland, R. (ed.) (2022) Oscar Wilde: The Complete Interviews, Little Eye, I, 448–54, 452) and I suspect that Constance would not have included the line unless she had heard her husband say it in the context of telling the story.
Constance is unlikely to have acted merely in a secretarial role, producing a fair copy of a story written by Oscar so that it could be submitted to a publisher. As Small notes, nineteenth-century printing houses typically destroyed author manuscripts once they had been set into type: ‘this circumstance is almost certainly responsible for the relative absence of textual witnesses to Wilde’s early fiction’.Small, I. (ed.) (2017) The Complete Works of Oscar Wilde, Vol. 8: The Short Fiction, Oxford University Press, xiii The implication must be that Constance’s draft preceded the final draft submitted to the publisher. We cannot know, however, whether it was the first draft, although the relative lack of emendations suggests that it was not.
If Oscar published a story under his own name that was partially or entirely the product of his wife’s pen, he would not be the only author to be guilty of such an act. Henry Gauthier-Villars famously published his wife Colette’s novels under his own pen name. And Zelda Fitzgerald, when reviewing her husband F. Scott’s 1922 novel The Beautiful and Damned, noticed that it included portions of her diary and commented that ‘Mr. Fitzgerald […] seems to believe that plagiarism begins at home’.