Two Purāṇas, one translator, one episode, and the endings are not the same ending.
In the Śiva Purāṇa, the search is answered with more work. Stop, and then go do austerities — the instruction points forward, to another effort, and the failed climb turns out to have been a preliminary.
In the Bhāgavata, the search is answered with a relocation. God does not tell him to try again. He names the failure back to him in plain terms and then says where the thing actually was: inside. The climb was not a preliminary to anything; it was looking in the wrong dimension, and the correction is not effort but direction.
That is a real doctrinal difference, sitting in one motif, and it is exactly the kind of difference a reader would want flagged.
The notes in range flag nothing. Note 843 reads, in full, "Brahma." Note 844 reads "Inside the lotus." Both are bare identifications. Debroy names the Śiva Purāṇa in this volume's Introduction only, in the list of eighteen and in the dating table, and there is no cross-reference to it anywhere near this passage. Why that is so turns out to matter, and the section below revises what this page originally concluded from it.
This is the local-gloss behaviour that the volume's headline finding documents inside one book, now visible between two books that the same person translated. [n≥3 PATTERN] holds for the intra-textual case. The cross-corpus case is a second instance of a different thing and is [n=2 SUGGESTIVE] at best — the "we have taken a few liberties" pair is the other one — and it is not being named as a signature here.
Originally filed with the claim that a translator working title-by-title never points across his own shelf, and the rule that any doctrinal difference between two Debroy volumes is an unflagged finding by default. The check was run the same day across all five of his volumes in this vault, and it refutes that flatly.
His endnotes point across constantly. In the Śiva Purāṇa alone: "The story is told in Bhagavata Purana" (Vol 1, n. 969, on Citraketu) · "narrated in Markandeya Purana" (Vol 2, n. 400) · "In the Valmiki Ramayana, Rishyamukha is to the north of the Vindhyas" (Vol 2, n. 845) · "According to Vishnu Purana…" (Vol 3, n. 1049) · "According to Bhagavata Purana, these eight types of devas are…" (Vol 3, n. 1158) · and two notes citing the Devī Bhāgavata (Vol 3, nn. 1461, 1498).
And one of them points directly at this build's own material. Śiva Purāṇa Vol 3, n. 1334: "Rudra Gita is from the Bhagavata Purana (4th Skandha) and was taught by Shiva to the Prachetas." That is Śiva Teaches Viṣṇu-Worship, 4(24), named and located from the other book.
The apparatus points in one direction only, and the direction is chronological.
Run the counts the other way and they collapse. The Bhāgavata Volume 1's endnotes name no other Purāṇa anywhere — not one, in 1,575 notes. Its only cross-title references are eleven to the Mahābhārata, which is not a Purāṇa and which he had already translated. Every other Purāṇa named in that volume appears in the Introduction's boilerplate — the list of eighteen, the guṇa taxonomy, the dating table.
The reason is mundane and is stated in his own apparatus: Śiva Purāṇa endnote 14 dates the Bhāgavata to Penguin 2018, and the Śiva Purāṇa's afterword describes choosing it as the next project afterwards. A 2018 book cannot cite a later one. The cross-references are not a policy about corpora; they are a translator reaching backward into what he has already done.
So the real finding is sharper than the one it replaces:
A growing translated corpus is asymmetrically annotated. Later volumes are dense with pointers backward; earlier volumes are sealed. Which means the oldest book on the shelf is the one whose apparatus is least aware of the shelf — and it is usually the one a reader starts with.
The stalk episode is a clean instance rather than a counter-example. The Bhāgavata's notes on it ("Brahma.", "Inside the lotus.") could not have pointed at a book that did not exist. And the Śiva Purāṇa's chapter 39, which could have pointed back, does not — its only Bhāgavata note is elsewhere and concerns Citraketu. So the connection is missable in both directions for two different reasons, one structural and one incidental.