The news we heard two or three weeks ago about the human remains identified as those of Richard III meant that several newscasters and commentators made use of the word skeletal. Those I heard on BBC R4 all pronounced it with penultimate stress, as skɪˈliːtl̩. There is, though, an alternative pronunciation, with initial stress, ˈskelɪtl̩.
So this is a word like palatal, in which uncertainty about the quantity of the penultimate vowel leads to rival pronunciations, one having the antepenultimate stressing and penultimate vowel reduction predicted if the penultimate is short (ˈpalətal, ˈskelɪtal), the other having the penultimate stressing to be expected if that vowel is long (paˈleɪtal, skeˈliːtal). In the case of palatal, phoneticians and linguists have settled for the first, although anatomists prefer the second (blog, 21 March 2011).
Since skeleton is derived from the Greek σκελετόν, where the epsilon spelling of the penultimate vowel indicates a short vowel, the regular pronunciation according to the Latin stress rule is indeed the usual one, ˈskelɪtn̩.
Actually, the adjective-forming suffix -al is normally disregarded for purposes of the stress rule: we keep the stress on the adjective in the same place as it has for the naked stem, thus from ˈperson we form ˈpersonal (despite the long ō in Latin persōna). From ˈuniverse we have ˌuniˈversal, because the extra syllable in the adjective means that Chomsky and Halle’s Alternating Stress Rule comes into play. So in skeletal, which does not involve an extra syllable compared with skeleton, it would be expected that the stress would be located on the same syllable as in skeleton.
And please don’t ask about adjectival, because I don’t understand why it’s ˌædʒɪkˈtaɪvl̩, either.