FIRST DAY. MONDAY-MORNING SESSION. The thirty-fourth annual convention of the Association of Official Agricultural Chemists was called to order by the President, J. K. Haywood of Washington, D. C., on the morning of November 19, 1917 at 10.15 at the New Willard, Washington, D. C. No referee on the subject of foods and feeding stuffs was appointed and no report on this subject was presented. No report on sugar in foods and feeding stuffs was made by the associate referee. REPORT ON CRUDE FIBER. By C. K. FRANCIS1 (Agricultural Experiment Station, Stillwater, Okla.), Associate Referee. A sample of wheat shorts and one of cottonseed meal, together with a piece of muslin for use in filtering, were sent to each collaborator. Determinations of crude fiber were made by the one filtration method and by the official method. The results are shown in the following table. Examination of the table shows that the results obtained by the one filtration method were considerably higher than those obtained by the official method. Moreover, results by the one filtration method are not so concordant as those by the official method. With the one filtration method it was particularly difficult for the collaborators to check each other. CONCLUSIONS. In the opinion of the collaborators, no saving in time is effected by using the one filtration method, and the failure of the results to check with those obtained by the official method raises a doubt as to whether the proposed method can be used with satisfaction. 1 Present address, Cosden & Company, Tulsa, Okla. By B. H. SILBERBERG (Bureau of Chemistry, Washington, D. C.), Associate Referee. The work during the past year on feed adulteration consisted of two parts: the first part on scratch feed based somewhat upon the recommendations in the 1916 report on feed adulteration; the second dealing with a method for the determination of hulls in cottonseed meal. PART I. The work on scratch feeds will be considered first. Two samples were prepared in the following manner and sent to each collaborator: The weed seeds and chaff were carefully cleaned out of some good quality commercial scratch feed which contained no grit. Each jar of both samples was then prepared separately. Three per cent of commercial grit and 2 per cent of a weed seed mixture were added to each jar of Sample 1. Six per cent of commercial grit and 8 per cent of a weed seed mixture were added to each jar of Sample 2. The weed seed mixture contained the following seeds, which represent large, mediumsized and small seeds: Agrostemma githago (corn cockle); Vaccaria vaccaria; Polygonum convolvulus (wild buckwheat); Chenopodium album (lamb's quarters). The collaborators were asked to make quantitative determinations of both grit and weed seeds, making two analyses if possible, using for one about 10 grams, and for the other about 20 grams. Later, through the work of A. J. Patten, one of the collaborators, and the results of experimentation, the fact was developed that the method used for mixing and drawing out the sample was of considerable importance. The samples were so clean and the ingredients for the most part of such size that the personal factor in making the separations was practically eliminated. The factors which would tend to cause a variation in results were thus narrowed down to two-the method of mixing and sampling, and the size of sample used-principally the former. The results obtained on this year's work would hardly justify drawing definite conclusions in regard to either of these points. However, for the time being, it would appear advisable to follow the recommendations of the former referee that, with a coarse scratch feed, no less than 20 grams be used for a sample, although if a thoroughly satisfactory method of sampling were devised, it might be possible to obtain accurate results on a smaller sample. As to the method of mixing and sampling, the results indicate that some mechanical method of mixing and sampling, in which the personal factor is reduced to a minimum, is practically a necessity for concordant results. Table 1 shows that in one instance an analyst obtained very good results by mixing on sampling cloth, reducing by quartering, etc., on a sample weighing only 9.8 grams. But by the same method of mixing and on a sample weighing about twice as much, the same analyst obtained results 25 per cent off on grit and 40.5 per cent off on weed seeds. In another case an analyst obtained correct results by this method of mixing on a sample weighing 37.0 grams, but the same analyst, on a sample weighing 19.8 grams, was 6.6 per cent off on grit and 15 per cent off on weed seeds, while on a sample weighing 10.38 grams, about the same amount as that on which the other analyst mentioned obtained very good results, his are 40.3 per cent off on grit and only 1 per cent off on weed seeds. One analyst who used the sampling cloth for mixing and sampling was 74.0 per cent off on weed seeds and grit combined on a sample weighing 15.26 grams, and another analyst using this method of mixing and a sample weighing slightly more (16.63 grams) was 91.2 per cent off. While these figures are taken from results on Sample 1, those on Sample 2 would lead to the same conclusion, that mixing and quartering on a sampling cloth is not a reliable method of drawing out a sample. Another method proposed is to separate the sample into three portions by passing it through 10 and 20 mesh sieves. This makes it easier to mix each portion more uniformly. Then, after mixing each of the three portions, weigh and take one-tenth of each for a sample. While this method has not been given so thorough a test as it would seem to merit, the results obtained so far indicate that, while it gives very satisfactory results in some instances, uniformly reliable results are not obtained. TABLE 1. Collaborative results on Sample 1, scratch feed, containing 3 per cent of grit and |