> Daniel JONES's has ["paeN °kres]. ([ae] = a + e ligature; [e] = schwa).
John Wells (1990) gives ["pæNkr&s] ([& = schwa]) as the only
possibility, both in <Pancras> and <St. Pancras>. I've always used this
pronunciation myself, blissfully unaware that I might run the risk of
being corrected by a London cabby :-). Well, I think the man was just
being hyperexplicit, trying to speak slowly and distinctly and using
what is sometimes known as 'foreigner talk':