Canto IV is a dynamic text that was changed by Pound and other editors well after its first publication date. These changes are still visible in the differences between the New Directions edition and the English text used in the Mondadori/Ugo Guanda bilingual edition of the Italian translations (Mary de Rachewiltz, 1985; Massimo Bacigalupo 2012).
The contemporary New Directions edition of The Cantos is heir to a massive effort of correcting errors in the poem, going back to Hugh Kenner and Eva Hesse’s activities of 1963-64 (Eastman 24). At that time, scholars aimed at reaching an agreement as to the definitive text of the poem and strove to correct errors in both the British Faber edition and the American New Directions one, editions which at that time varied from each other. However, authorization was an uncertain and tortuous road. Pound’s health was failing and from some point in time on, he ceased to care. The bulk of corrections was made in two stages, for the New Directions text of 1970 (1-84) and 1971 (85-117). Pound also wanted alterations and additions made, which never found their way into the New Directions text as we now have it. In 1975, Faber decided to scrap their own edition and take over the sheets from New Directions. Since that date, the UK and US editions have been identical. At the same time, James Laughlin, in consultation with Hugh Kenner decided to stop revising the text of the poem (Eastman 28).
Yet, the old Faber edition has survived, modified by Mary de Rachewiltz according to her father’s last corrections after his return to Italy in 1958. The root of this development is the bilingual Lerici edition of A Draft of XXX Cantos, published in 1961, also known as I primi trenta, which Hugh Kenner acknowledged as a prime source for the cantos it contains (Kenner xix). Mary de Rachewiltz, the first translator of The Cantos into Italian, worked on her translation with Pound’s assistance and declared that the poet included last-minute changes into the Faber copy of The Cantos she was using. These changes were incorporated into the Lerici edition and later flowed into the Mondadori text of 1985 and from there into various reprintings of the English text in Italy. Thus, the Mondadori is the survival of the Faber text of 1954 and 1964; it contains the last revisions that Pound made while re-reading The Cantos at Brunnenburg in 1958-59 and subsequent occasional modifications for later cantos made by Mary de Rachewiltz and detailed in her “Avvertenza” (I Cantos, 2013, 41-43).
There are two major instances in Canto IV where variants are evident and significant. See also the correspondence about them included in the Calendar (Taylor 2002, 2015).
1. lines 86-87:
Saffron sandal so petals the narrow foot: Hymenæus Io!
Hymen, Io Hymenæe! Aurunculeia!
Saffron sandal so petals the narrow foot: Ύμήν,
Ύμέναι ώ, Aurunculeia! Ύμήν, Ύμέναι ώ,
2. lines 89-93
And So-Gyoku, saying:
"This wind, sire, is the king's wind,
This wind is wind of the palace,
Shaking imperial water-jets."
And Hsiang, opening his collar:
“this wind, sire, is the king’s wind,
This wind is wind of the palace,
Shaking imperial water-jets.”
And Ran-ti opened his collar:
Maybe the last word in the matter should be left to Hugh Kenner, who summed up the editorial situation:
"The Cantos are a corrector's paradise. There are printers' errors. There are discrepancies between parallel printings, sometimes but not always ascribable to a printer. There are errors of fact or transcription committed by the author [...] 'Ran-ti' [...] is a place Pound mistook for a king in the Fenollosa notes, but the 'Hsiang' that replaced it in the New Directions text of 1970 is a Chinese name that clashes with that of the Japanese interlocutor So-Gyoku, and unless it's the author's correction I'd uncorrect it. (Pound did make several alterations in this passage which he wrote into my copy in 1950 and had Faber effect in their Seventy Cantos of that year, but he left the names alone.)
And so on. It's easy to go on tinkering forever, sometimes missing the point, sometimes spoiling the sound for the sake of obtaining accord with an encyclopedia, sometimes imposing a consistency the author had avoided in his effort to reflect his source of the moment, not his own usage of a hundred pages back.
And yet, there are errors to correct, and when New Directions finally adopted the offset process in 1970 for the first complete edition (Cantos 1-117), they not only (finally!) paginated the book straight through, they made a bold start at emending. 'Veil' became 'viel,' the unpronounceable Greek was rectified, all manner of minutiae got set straight. But that was also when 'Ran-Ti' became 'Hsiang,' a change of a different order because the error had been the author's, and had entailed moreover not a slip of the pen but substantive syllables he let go through his head many times when he balanced the sound of the passage in 1919 and again when he rebalanced it in 1950 after he'd changed five Latin words to six Greek ones. Though there can be no doubt that it originated in a misreading, we must take 'Ran-Ti' as embedded in the passage, to be footnoted but not altered.
Once you start trying to help the author get things right, there's no stopping till you've rewritten the entire poem to new criteria, bollixing in the process criteria of Pound's (So-Gyoku and Hsiang won't co-exist in the same linguistic or poetic universe).
Which is not to blame James Laughlin and Fred Martin at New Directions. Unlike the decision-makers at Faber & Faber, whose salutary conservatism was buttressed by a suspicion that the poem made no sense anyway and also sold quite well the way it was, the Americans wanted the Cantos to show Pound's best foot put forward. But they were inundated with advice, and the advisors (I was one) were imperfectly in touch with one another; James Laughlin wrote me in 1963 that he received approximately a letter a week from someone with a correction to the Cantos. Under this pressure, and with the printer waiting, they were compelled to assume a role for which God had not intended any publisher, that of textual editor, choosing which corrections to implement, moreover without the leisure to formulate principles. During three printings (1970-1973) they made 232-too many. And records are adrift, and it's no longer easy to say where most of the changes came from."
Hugh Kenner. "Introduction." In Barbara Eastman. Ezra Pound's Cantos: the Story of the Text 1948-1975. Orono: Maine: National Poetry Foundation, 1979. xviii-xix.