Permeable and impermeable seeds[4] of _Melilotus alba_ and _M. officinalis_ were obtained from commercial samples and also from samples collected in the field. Those selected for sectioning were allowed to dry after being removed from the germinator and then embedded on the ends of pine blocks in glycerin gum, which was made by dissolving 10 grams of powdered gum arabic in 10 c. c. of water and adding 40 drops of glycerin. After the glycerin gum had dried for 24 hours, the seeds were easily sectioned. This method of embedding causes no change in the seed coat. It is more satisfactory than the paraffin method for holding the seeds firmly. The glycerin gum dissolved readily when the sections were mounted in water. [4] The term "permeable" is used in this paper to designate seeds whose coats are permeable to water in two weeks or less at temperatures favorable for germination, while the term "impermeable" is used to designate seeds whose seed coats are impermeable to water for this length of time when temperatures are favorable for germination. Impermeable seeds are commonly referred to as "hard seeds," and they may become permeable in time. In the microchemical studies Sudan III, alcanin, chlorophyll solution, and phosphoric acid iodin were used to test for cutin or suberin; sulphuric acid and iodin, chloriodid of zinc, and chloriodid of calcium for cellulose; phloroglucin and hydrochloric acid for lignin; ruthenium red for pectic substances; and sulphuric acid, Congo red, and aniline blue for callose. Where very thin sections were necessary for detailed study of the structure of the seed coat, pods in various stages of development were collected, and after the usual preliminary treatment they were embedded in paraffin and sectioned with the microtome. Microchemical tests were made with these sections by using various specific stains. Safranin was used to test for cutin, suberin, and lignin; haematoxylin and methyl blue for cellulose ; methylene blue, methyl violet B, mauvein, and ruthenium red for pectic substances; and aniline blue and Congo red for callose. In studying some points with reference to the pore system of the seed coat, it was necessary to use free-hand sections of fresh pods. In studying the seed coat in relation to the absorption of water, both permeable and impermeable seeds were soaked in water solutions of safranin, gentian violet, eosin, and haematoxylin, then dried and embedded in glycerin gum for sectioning. Seeds were soaked in stains dissolved in 95 per cent alcohol to test the penetration of alcohol. It was evident that the seed coats did not act as a filter, as the stains passed through them with the water or alcohol. STRUCTURE OF THE SEED COAT. There is very little endosperm present in mature seeds of _Melilotus alba_ or _M. officinalis_. That which is present is quite permeable to water and therefore bears no relation to the impermeable seeds of these plants.