What is uncle block?

A valid block that arrived too late to become part of the main chain, which Ethereum's proof-of-work design rewarded partially instead of discarding.

Also called an ommer, since not every such block is an uncle in the strict family sense. The mechanism is historical: it belonged to Ethereum before the Merge and does not exist under proof of stake. You still meet the term in old block data, in explorers that display an ommers field, and in explanations of why Ethereum's block times could be short.

The thing to understand is why a chain would pay for blocks it does not use.

How it works

Two miners can solve a block at nearly the same moment. Only one can extend the chain, and under Bitcoin's rules the other is orphaned and its miner is paid nothing. Bitcoin's roughly ten-minute target makes this rare.

Ethereum targeted a block time in the low teens of seconds, so near-simultaneous blocks were common, and orphaning them would have systematically favored miners with the fastest network connections and the largest pools. Rewarding them instead removed most of that advantage.

The Ethereum Yellow Paper specifies the rules. A block could include at most two ommers, each no more than six blocks behind the current one. The ommer's miner received a reward scaled by how stale it was, computed as the static block reward multiplied by eight plus the ommer's number minus the including block's number, divided by eight. The miner who included it received an additional one thirty-second of a block reward for doing so.

The Merge ended all of it. Proof of stake assigns exactly one proposer per slot, so there is no race to produce a competing block and nothing to compensate. A slot whose proposer misses it is simply empty.

Example

Illustrative under the old rules with a 2 ETH static block reward. Block 1,000,000 is mined. A competing valid block at height 1,000,000 arrives moments too late and is included as an ommer in block 1,000,001.

The staleness distance is one, so the ommer reward is 2 multiplied by (8 plus 1,000,000 minus 1,000,001) divided by 8, which is 2 times 7 divided by 8, or 1.75 ETH. The miner of block 1,000,001 receives an extra 2 divided by 32, or 0.0625 ETH, for including it. Had the ommer been six blocks behind, the fraction would have fallen to two eighths and the reward to 0.5 ETH.

Why it matters when you buy

Uncle blocks are not something a buyer encounters today, and the reason they mattered still is. Short block times mean a chain produces more near-simultaneous blocks and more shallow reorganizations, which is why exchanges set different confirmation counts per network before crediting a deposit. The chain pages cover settlement behavior by network.

Related terms

• the merge: when the mechanism ended
• slot: proof of stake's one-proposer-per-slot replacement
• chain reorganization: what competing blocks cause
• block reward: the payment ommers were scaled from
• proof of work: the consensus this belonged to
• confirmation: why competing blocks matter to deposits

Questions

Do uncle blocks still happen on Ethereum?

No. Proof of stake schedules one proposer per slot, so there is no competition to produce a block at a given height and nothing to reward as an ommer.

Is an uncle block the same as an orphan block?

Not quite. An orphan block is discarded and pays nothing, which is Bitcoin's treatment. An uncle was referenced by a later block and paid a reduced reward, which was Ethereum's.

Did uncles change Ethereum's total supply?

Yes, they added issuance beyond the canonical chain's block rewards. That is one reason historical issuance figures for Ethereum's proof-of-work era are higher than a simple block count would suggest.