Here's what Michael Quinion has to say on the subject: [Q] From William Murray; a related question came from Tim Riley in London: "The title of your new book, Ologies and Isms, reminded me of a thought I've had about kilometer. The word itself is not essential in the USA, except for scientific measurements. I use two different pronunciations depending on context: in scientific and most contexts, I'll use KILometers, but for road distances (and common measure when I'm in Canada), I'll say kiLOMeters. I wonder if this dual pronunciation nicety is common in the rest of the English-speaking world."
[A] The version with the stress on the second syllable used to be rather more common in the US, with the rest of the English-speaking world tending to stress the word on the first syllable. In Britain and the Commonwealth the one stressing the first syllable is still considered by some to be the only correct form. But the situation is changing. I hear it stressed on the second syllable quite regularly from reporters on the BBC news, for example. Professor John Wells of London University surveyed British usage for the 1998 edition of his Longman Pronunciation Dictionary and found that in the decade since his last survey the form with initial stress had lost ground to the other. In 1988 kilometre with initial stress was just in the majority, but in ten years had slipped back to only 43% of those who replied. If that trend continues, it will take only a generation for the non-specialist American pronunciation to take over entirely in Britain. It's ironic that the American pronunciation should prevail when Americans-almost the last users of non-metric measures in the world-have less need for the word than other English speakers. As you say, in scientific use the stress tends to be more often on the first syllable, perhaps because the significance of the prefix is more obvious in technical contexts. What puzzles people about this stress shift with kilometre is that it doesn' t occur with most other metric units in -metre (-meter if you're American): we don't move it in centimetre or millimetre, for example. Also, it doesn't happen with other units prefixed with kilo-, such as kilogram and kilolitre (the first of these is unlikely because it has too few syllables, but the second ought to be possible). It does happen sometimes with micrometre, however, which can take on the same stress as the measuring instrument, which is spelled micrometer everywhere. This gives us the clue to what is happening-it's by analogy. Every measuring instrument whose name ends in -ometer has the stress on the antepenultimate syllable, the one containing the o: thermometer, barometer, chronometer, hygrometer, speedometer, galvanometer, sphygmomanometer, and so on. The stress on kilometre is merely following the trend. The American spelling as kilometer has possibly reinforced the connection. This, of course, only pushes the problem back one step. Why should all these measuring instruments have their names said this way? It seems in part to be linked to the concept of strong and weak vowels in English. Some, such as i and u, are usually weak, while the remainder, including o, are usually strong. All that these terms mean in practice is that syllables containing strong vowels (such as the o in -ometer) are more likely to receive the stress. There is also a tendency in compounds for the stress to move towards the end of the word. But these are only tentative suggestions, since the history of language is full of unpredictable shifts in stress. Professor Wells pointed out that various historical studies "show that many words have fluctuated wildly in their stress patterns over the years. Why did we change from cha-RAC-ter to CHAR-acter, or from bal-CO-ny to BAL-cony? I don't think anyone really knows".
