take a look at:

block hashes:
0x00000000000271A2DC26E7667F8419F2E15416DC6955E5A6C6CDF3F2574DD08E
0x00000000000743F190A18C5577A3C2D2A1F610AE9601AC046A38084CCB7CD721
they have duplicate coinbase transactions:
0xE3BF3D07D4B0375638D5F1DB5255FE07BA2C4CB067CD81B84EE974B6585FB468
block height 91722, 91880, on 2010-11-14 08:37:28, 2010-11-15 00:36:19

block hashes:
0x00000000000AF0AED4792B1ACEE3D966AF36CF5DEF14935DB8DE83D6F9306F2F
0x00000000000A4D0A398161FFC163C503763B1F4360639393E0E4C8E300E0CAEC
they have duplicate coinbase transactions:
0xD5D27987D2A3DFC724E359870C6644B40E497BDC0589A033220FE15429D88599
block height 91812, 91842, on 2010-11-14 17:59:48, 2010-11-14 21:04:51

The miner was using the same signature "0x0456720E1B00" on all 4 of these coinbase transactions.
The public keys for those two pairs of duplicates were:
"0x4104124B212F5416598A92CCEC88819105179DCB2550D571842601492718273FE
0F2179A9695096BFF94CD99DCCCDEA7CD9BD943BFCA8FEA649CAC963411979A33E9AC"
and
"0x41046896ECFC449CB8560594EB7F413F199DEB9B4E5D947A142E7DC7D2DE0B811
B8E204833EA2A2FD9D4C7B153A8CA7661D0A0B7FC981DF1F42F55D64B26B3DA1E9CAC"

• Should we allow this guy to spend his duplicate transactions? This would make code and data structures more complex for eternity... when verifying transactions we'd have to get a list of transactions for each hash, instead of just a single transaction... and mark a non-spent one as newly spent.
• Or should we make a rules like "no more duplicate transactions when height > 91880" and "only one of a set of existing duplicate transactions are spendable coinbase"? I like this. :) Come on, we totally should have rejected those duplicate transactions!