It was communicating with the terminal program at 9600 baud without a problem... wouldn't it have spouted gibberish if it didn't use 8Mhz? I'll use uppercase to be safe.

But I think I figured out the problem... I was communicating with the clock pin idling low... it should have been idling high. Therefore the pic was not shifting the right data or the right register address to the ADXL375, and so it never got into measurement mode. After I changed it to idle high, I started getting values that looked meaningful (but very noisy!), although I think that's just because I don't have any filtering on the power supply or at the IC pins (just wanted to get it working). I'll try that next.

Dave