As Williams himself points out the method is far from ideal for general use since some practice is necessary before it can be applied with any confidence, and a personal factor is introduced to quite an extent. Williams9 gravimetric method, which he considers preferable, is carried out as follows: *7

The medium contains 20.0 gm. of cane sugar. 3.0 " " ammonium sulphate 2.0 " " potassium dihydrogen phosphate.

1.5 gm. of asparagine 025 " " calcium chlorid 0.25 " " magnesium sulphate.

All dissolved in 1 liter of distilled water.

One hundred cc. of this solution are put into a 500 cc. Erlenmeyer flask and a known amount of the solution to be tested is added. The volume is made up to 110 cc, the flask is plugged with cotton, sterilized or pasteurized to kill all vegetable organisms, cooled, and brought to a temperature of 30° C. in an incubator.

A yeast suspension is made by weighing out 0.300 gm. of fresh Fleischmann's yeast (small cake in tin-foil) taken from the center of the cake; this is made into a paste with a very small amount of water and suspended finally in 1 liter of sterile water. 1 cc. of this suspension well shaken and freshly made is introduced into the culture medium with a sterile pipette. By this means 0.3 mg. of yeast is used for seeding.

The flasks are then put into the incubator and the yeast is allowed to grow undisturbed 18 hours at 30° C, when the growth is stopped by the addition of a little formaldehyde solution. If the seed yeast has not been kept in a refrigerator continuously since manufacture, a very rapidly growing wild yeast may appear to a slight extent, floating on the surface of the solution. This should be removed by lifting it from the surface with a small piece of fine copper gauze attached at right angles on the end of a glass rod or tube. This removes practically none of the other yeast, most of which settles to the bottom.

The yeast is then filtered off on a weighed " Gooch " crucible, prepared with paper underneath the asbestos, washed thoroughly with water, finally with a little alcohol, dried for 2 hours at 103° C, and weighed after an hour's cooling. The "Gooch " crucibles cannot be used repeatedly without treatment as the dried yeast loses weight on washing with water.

57 Williams, J. Biol Chem. 42, 259, 1820. See also Swoboda, Ibid, 44, 531,1920.

If no vitamin is added to the solution, the yeast produced will be about 2.5 mg. The amount of growth above that of the blank is, within limits, directly proportional to the amount of vitamin put into the solution. The results may be expressed numerically and the number represents the relative purity of the preparation tested or the value of the material tested as a source of this vitamin. The "vitamin number" of a material may be defined as the number of milligrams of yeast produced by the addition of its extract, minus that produced in a control solution, under given conditions and within certain limits, computed to 1 gm. of the original material tested.

Of this procedure Williams observes that the vitamin to be tested must be in solution. No method of extraction is outlined. According to Williams nothing except vitamin (especially in small amounts) is able to produce more rapid growth, hence the vitamin does not need to be prepared especially free from other materials, as impurities in all probability will have no accelerating effect. The total amount of material added ordinarily will be much less than 0.1 gm. In using this method for vitamin determinations, precautions must be taken that the vitamin is extracted quantitatively from the material in question. Investigators have found that this is difficult to do, especially from some materials. In the handling of the extracts care must be taken that toxic substances are not introduced or formed which will counteract the effect of the vitamin which may be present. This is true also of animal feeding experiments where poisonous substances must be avoided. Treatment of material with acid especially at high temperature or overheating in the dry condition frequently produces something toxic in yeast, while treatment with alkali may destroy the vitamin. In some extracts the vitamin is destroyed to an appreciable extent if kept at the boiling temperature an hour or more.

An autolysate of yeast is not a desirable material as a yeast nutrient as inhibiting substances appear to be present. If the amount of growth produced is not proportional to the material added, but falls off with the larger quantity, it is good evidence that something toxid is present, which becomes more potent in the higher concentration and counteracts the effect of the increase in vitamin. This can be confirmed by microscopic examination of the yeast produced, which reveals small granular irregular cells or other abnormalities. It is mainly due to the possible presence of toxic substances that two dilutions should be used for a determination. If toxic substances are present the results from the lower concentrations are more reliable.

A comparison of the vitamin content of various foodstuffs as determined by Williams with yeast and by Osborne and Mendel with rats gave for the most part fairly concordant results.58 Where discrepancies existed there are some indications that vitamin C might be acting as a second factor in stimulating yeast growth. The yeast test indicates a higher vitamin content in bakers' yeast than in brewers' yeast, whereas the animal tests indicate the reverse. Yeast extracts appear, however, to exert a specific influence on different varieties of yeast, since extract of brewers' yeast was found to be a better stimulant for the growth of brewers' yeast than bakers' yeast extract, while the latter had more influence on the growth of bakers' yeast.

Eddy and Stevenson59 experimented with both the Bachmann and the first Williams methods of quantitative study of vitamin content, and finally devised a modification of both which they report as satisfactory.