This study introduces and analyzes a novel authorship attribution case: the children’s stories published by Oscar and Constance Wilde. We analyzed the corpus of stories with both supervised (SVM with string kernel) and unsupervised (Hierarchical Clustering via Rank Distance) methods and found a strong stylistic similarity between the story "The Selfish Giant" published by Oscar Wilde and the stylometric profile of Constance Wilde. Starting from this baseline, we also explored the the capabilities of LLMs in authorship attribution via Perplexity. Our finding suggests that the story "The Selfish Giant" might be the result of a collaboration between Oscar and Constance Wilde. Moreover, our results pointed to the distinct stylistic fingerprints of the two authors with regards to the rest of the corpus, confirming that their respective styles are separable despite shared genre and period.