^{page 189 note 1} A reprint of M. Bonneau's article has been published by Librairie Paul Geuthner (Paris, 1938).Google Scholar I have to thank my former colleague Mr. James Liu (now at Hong Kong University) for bringing Mr. Fraser's article to my notice and for giving me a copy of the magazine containing it.

^{page 192 note 1} It should perhaps be explained that tatte is being treated as [tat:te], two syllables and six phones; and that the following combinations of letters in Japanese are being regarded as diphthongs, ae, ai, ao (but not in masurao), au, ou, and ui in uguisu (= wii).

^{page 199 note 1} sic. ? insert ‘it is hard to know’ before ‘what’.

^{page 199 note 2} By Otome Daniels. The English version can perhaps be said to conform to M. Bonneau's rule.

^{page 200 note 1} It may be that this type of difficulty so severely restricts the use of the method that it could never be used in any situation demanding that a large proportion, if not the whole, of some pre-selected body of verse should be translated. But one translator might succeed where another would fail, and it is hard to tell in what proportion of cases a difficulty of this type would really be insuperable.

Where absolutely necessary, some relaxation might be authorized. A line with SP within 10 per cent, instead of 5 per cent, of the SP of its original might be allowed. Or, where the Japanese line is not ‘full’—where, e.g., though it has five kana units, it has less than five syllables or less than ten phones—an English line might be allowed to match the SP of a ‘full’ line with the same number of kana units. Either of these relaxations would allow ‘The ancient pond’ as the first line of the translation in our example. The latter relaxation would give no relief if the line was already ‘full’; to meet such cases some relaxation in a downward direction might be necessary.