It seems like there's some variation in the sensitivity of different meters to IR. I've had some luck metering through an R72, but for some people it's been a complete failure; I don't know if the difference is down to different built-in filtration, or different lighting conditions, or just blind luck...

Even guidelines based on assuming a certain EI are variable, because the ratio of visible light to IR light varies. You know how sunsets look red---that's because red light propagates better than other colors through the atmosphere and haze and dust, and the same goes for infrared. As a result, at sunrise or sunset, the EI of an IR film is *higher* than it is at noon! (There's less total light, but what light there is skews to the red/infrared---thus IR film through a filter actually sees more light at those times than the amount of visible light would suggest.)

While metering through the filter might work in a pinch, on the whole I wouldn't count on it. I tend to think it's always best to shoot IR film with a handheld meter---and then the hard part is choosing what speed (EI) to set the meter to. The speed that's worked for other people is a starting point, but for your particular conditions it may or may not be ideal. I'd say, take someone else's suggested EI as a start, shoot an experimental roll with some bracketing, and evaluate the results to figure out what works for you.

A compensating developer is also your friend. I've become very fond of IR films in Diafine, which seems to "level out" the exposure variability quite a bit.

-NT