100uF was what I had laying around. Actually I think the positive rail got 220u and the negative rail 100u, or vice versa. The data sheet for it (uA7915 from TI) explicitly said no extra components required (though I admit I almost always use them anyway). It's an obselete part, but I thought it odd that the entire tube of them I had showed the same thing contrary to the data sheet.
ETA: I found the application note that made me think that they weren't necessary on standard (NPN darlington pass transistor) regulators, see pg19:
**broken link removed**
It uses the word 'usually'. Guess I found the non-usual case