1. Agar-agar after suitable preliminary treatment can be acetylated easily with pyridine and acetic anhydride. 2. The Agar acetate obtained can be simultaneously deacetylated and methylated to give methylated agar 3. Hydrolysis of the methylated agar with dilute mineral acid, followed by esterification, gives methyl laevulate, 2:4:6-trimethyl methylcgalactosid and a substance behaving as a methylated ketoside 4. The 2:4:6-trimethyl methyl^galactoside on hydrolysis gives 2:4:6-trimethyl a-d-galactose, the structure of which is proved by (a) Complete methylation which gives 2:3-4:6-tetramethyl galactose. (b) The isolation of a crystalline dimethyl galactose phenyl osazone. (c) Oxidation with bromine water forming a trimethl-galactonolactone, which gave a crystalline amide. (d) Oxidation with nitric acid, followed by esterification and amide formation which gave product of indefinite composition, neither trimethyl mucic acid nor dimethoxy succinic acid being obtained. It would appear therefore that the galactose units are of the ß-form and linked at positions 1 and 3. Methylated agar can also be hydrolysed with 1.2% methyl alcoholic hydrogen chloride to give a trace of methyl laevulate, 2:4:6-trimethyl methyl- galactoside and the substance behaving as a methylated ketose. 7. Extensive examination of the latter fraction has been hindered for lack of material. Complete methylation yielded a crystalline "Trimethyl anhydr ketose" C₉H₁₆O₅ but oxidative degradation has so far been unsuccessful in determining the structure of this fragment. The original "dimethyl methyl≃ketoside" appears to be a mixture but oxidation with nitric acid failed to produce any dimethoxy succinii acid which is taken as evidence that the two rnethoxyl residues apparently present are not adjacent.