The Visuddhimagga was edited and then published twice in Roman script in the first half of the 20th century.  By Caroline A. F. Rhys Davids for the PTS, published 1920 & 1921, and by Henry Clarke Warren for HOS, posthumously published in 1950.  Neither edition refers to the other. 

Did Warren not know about Caroline Rhys Davids' work?  Warren met him in Oxford in 1884, as Lanman's Memorial notes, and was greatly influenced by him.  Warren's work was done long  before that of Caroline Rhys Davids, since Warren died in 1899.  But why wouldn't Dharmananda Kosambi have mentioned Caroline RD's edition in his 1927 preface to Warren's?  It's understandable that Warren's brother Edward wouldn't have known about Caroline RD's edition, when he wrote his pathetic Foreword in 1927, since he was not an indologist.  Why did Warren's edition take 23 years to be printed, even after Kosambi had finished his editing of the MS?  1950 looks like five years after the war, which is understandable.  But that doesn't explain the twelve years of inaction before the war (and after the editing).  Since Warren had paid for the HOS to exist, one would have thought some priority might have been given to publishing his work.

And why didn't Caroline RD mention Warren's work?  Warren had used one of her husband's manuscripts of the VM, so there would surely have been some awareness of Warren's work.  And Thomas Rhys Davids was alive until the end of 1922, and was aware of his wife's work on the VM, since she gave him some pages for checking, some time before the end of 1920 (mentioned in her foreword).  Caroline RD also knew that Warren had published an subject analysis of the VM in the JPTS in 1892, but appears not to know his article "Buddhaghosa's VM" of the same year, or his "Report of Progress" on his work on the VM, published in 1894.   She mentions Warren in her afterword on p. 767, but only as the author of Buddh. in Tr. (1896), which incidentally contains a 50 passages  translated from the VM. 

Has someone worked out all these matters?  I would have expected Ñanamoli to say something about this in his translation, but he doesn't.  He just says he's using both editions.  Perhaps there are book reviews from the later 1920s or 1950s that explain matters, I haven't looked yet.

Has anyone systematically compared the two editions for variants?  Caroline RD's edition is a reproduction of four earlier printed editions, two from Rangoon, two from Ceylon; Warren worked from MSS, two Burmese and two from Ceylon.  One of Warrens' MSS came from the India Office and one from T RD's private collection, as mentioned, so it must have been known in England, and to T RD, that he was working on this text.

Best,
Dominik