Diazomethane rapidly converts 3-acetylcoumarin, 3-cyanocoumarin, and their derivatives into the related 4-methyl-coumarins. Since an acetyl group must rotate out of the plane of conjugation to allow the methyl group to enter, it is a less effective activating substituent than cyanide in alkylation reactions.

Substituents at the 5-position interfere with the methylation process so that 3-acetyl-5,7-dimethylcoumarin adds diazomethane giving a pyrazoline derivative (XIII) in the usual way. Methanol converts this pyrazoline into oxepin lactones, e.g. 3-acetyl-4,5-dihydro-4-methoxy-6,8-dimethyl-1-benzoxepin-2(3H)-one (XVIII). Thermolysis of the pyrazoline is unusual in regenerating some of the coumarin from which it was made; the main product, however, is a labile substance giving 3-acetyl-6,8-dimethyl-1-benzoxepin-2(3H)-one (XVII) when isolation was attempted

At –40° diazomethane and 3-acetyl-5,6-benzocoumarin give 1,11c-dihydro-3a-(2-methyloxiranyl)benzo-[5,6]chromeno[3,4-c]pyrazol-4(3aH)-one (XXIV), but at 0° the sole isolable product is 8a-acetyl-11a,12-dihydro-11H-naphtho[1′,2′:6,7]oxepino[3,4-c]pyrazol-8(8aH)-one (XXII).

These results support the view that the ‘direct’ methylation begins with an intermediate or transition state having the shape of a pyrazoline but at least one bond which is as much ionic as covalent in character. It continues with hydrogen migration because this restores the delocalisation energy of the coumarin system lost during the addition. The arguments also clarify some of the reasons for cyclopropane ring formation in certain related series.