I noticed the author of that code mentioned using a 2 mega ohm resistor instead of the 10M speced in the data sheet. I just swapped that and get readings of around -41.
Here's where it gets interesting. I dug up an old Radio Shack temp/humidity gauge I had buried with my beer brewing equipment. It's reading 36% sitting on my desk, which is pretty dam close to what the Humirel is saying (after ignoring the sign issue). If I breathe into it a few times it jumps to around 84% which is a little more than the Humirel jumped but still in the ballpark.
Not sure if I should leave his formula as is and experiment with different resistor values (not the pulldown, the RC) or tweak his formula:
#define RH1(time) ((.1667*time)-67)
...based on some extrapolation. Or maybe try to mess around with the formula in the data sheet instead: (Time - RHConstant) / 24 where RHConstant is 12169 (though not sure what that is based on and may be different for Arduino).