Quite so. A little more digging made me come across the OFT leaflet which gives the statutory formula for calculating APR.
Having given the actual formula, it then states the following:
" Although it is uncommon, complex mathematical equations such as the APR calculation can sometimes have more than one correct answer (there could be more than one value of i at which the two sides of the statutory equation are equal). The Regulations set down the rules for dealing with this:
if there is more than one positive result, or positive and negative results, the APR is the positive rate nearest to zero;
if there is no positive result, the APR is the negative rate nearest to zero. "
Good, eh? Sometimes they really do think of everything. I was surprised.