three million dollars for mixed and single element commercial fertilizers, considerable of which comes from the Pacific coast (San Francisco chiefly). Fine soil is sometimes used as a "filler" in these "mixed goods" and, as this is never sterilized, it is conceivable that it might carry Azotobacter species. Dried blood, guano and tankage are also shipped here in considerable quantity, which may also be contaminated with these microorganisms. During recent years, green manuring with legumes has been quite extensively practiced on several of the larger plantations. These legume seeds have been imported from various parts of the world and might occasionally harbor Azotobacter. Be that as it may, we are sure of one fact: Azotobacter are quite universally distributed in Hawaiian soils. SUMMARY 1. Thirty soils from different localities on the four large islands of the Hawaiian Group were examined as to their Azotobacter content. Their abilities to fix nitrogen in mannite solutions are given. 2. Only five soils were noted which failed to show Azotobacter growth in solutions. 3. Four different forms of Azotobacter were isolated in pure cultures, described, and their abilities to fix nitrogen in solutions and in soils given. 4. A possible explanation of the introduction and distribution of Azotobacter in these isolated island soils is advanced. LITERATURE CITED (1) LIPMAN, C. B., and BURGESS, P. S. 1915. Studies on nitrogen fixation and Azotobacter forms in soils of foreign countries. In Centbl. Bakt. (etc.), Abt. 2, Bd. 44, p. 481-511. (2) LIPMAN, C. B., and PRESSEY, H. F. 1913. A contribution to our methods of determining nitrogen in humus. In Jour. Indus. Engin. Chem., v. 5, p. 143-144. THE EFFECT OF TIME AND DEPTH OF CULTIVATING A WHEAT SEED-BED UPON BACTERIAL ACTIVITY IN THE SOIL' By P. L. GAINEY, Soil Bacteriologist, Kansas Agricultural Experiment Station Before the American Society of Agronomy 1914, L. E. Call (1) of the Kansas Agricultural Experiment Station, presented a very interesting paper concerning the effects that different methods of preparing a seedbed for wheat have upon yield, soil moisture, and nitrates. The principal fact with which we are concerned, brought out in this paper, is the very marked effects that the various experimental methods have upon the accumulation of nitrates. So striking and consistent have the results been for the past several years that a careful study of the problem was deemed advisable. A study of the bacteriological factors involved was given the writer. The project, as conducted by the Department of Agronomy, consists in comparing eleven different methods of preparing seed beds. A detailed description of the various methods, with results secured, etc., is given in the above mentioned paper. The major points of comparison have been time of preparation and the depth of culture. The dates of preparation vary from July 15, to seeding time, about October 1. The depth of culture varies from discing to plowing 7 inches deep. In general, the effect upon nitrate accumulation has been that early and deep preparations give materially higher nitrate contents at seeding time than do late and shallow. A few specific examples from Call's paper will suffice to show the influence of these two factors. 22.43 lbs. NO, per acre. 57.30 lbs. NO, per acre 517.01 lbs. NO, per acre. 76.83 lbs. NO, per acre. Plot No. 1 disced 1 Contribution from the Research Laboratory in Soil Biology, Kansas Agricultural Experiment Station. An abstract of this paper was read before the Society of American Bacteriologists, 1915 meeting. Received for publication May 22, 1916. From these data it is evident that some very important factors must be influencing the nitrate accumulation in the various plots. For the bacteriological investigations it was deemed advisable to restrict the studies to a very limited number of plots. If the factors operating in the plots studied could be ascertained an application of similar methods to the whole series would be undertaken, provided such seemed advisable. As representative of the extremes, both in methods of preparation and in results secured, so far as yield and nitrates are concerned, we chose as plots upon which to work Nos. 1 and 9. The treatment of No. 1 has been thorough discing at seeding time (about Oct. 1). The treatment of No. 9 has been plowing 7 inches deep July 15, and cultivating as often thereafter as necessary to insure a good seed-bed. For the four seasons '11, '12, '13 and '15, No. 1 has produced an average of 6.9 bushels of wheat per acre, while No. 9 has produced 27.59 bushels. The nitrate content for the four preceding seasons at the last analysis was for No. 1, 42.95 pounds of NO, per acre, and for No. 9, 318.69 pounds. From these data it is evident either that there has been much more NO, formed in No. 9, or that much has disappeared from No. 1. The only evident source of loss existing in No. 1 and not present in No. 9, is in the growth of weeds taking place prior to cultivation. Analysis has shown that in extreme cases this can account for only a small portion of the observed differences. It would seem then that those plots showing high NO, content actually form larger quantities. We have, therefore, for two seasons been directing our attention toward the nitrate forming power of soil from the two plots under study. EXPERIMENTAL DATA The experiments given below include only a very few of those that have been carried out. An effort has been made to select representative experiments. At a later date a more complete report will be published. AMMONIA FORMING POWER The relative ability of soil, from the two plots under study, to liberate ammonia from organic nitrogenous compounds is shown in Table I. Soil for the experiments here given was collected by means of a 2-inch soil auger. Twelve cores to a depth of 12 inches were drawn from each plot. Care was used to prevent, as far as possible, outside contamination. The soil was brought to the laboratory, passed through a 3-mm, sieve, and thoroughly mixed. The moisture content and water holding capacity were determined. Four samples, the equivalent of 100 gm. dry soil, were then weighed. To two each of these cottonseed meal and dried-blood containing 60 mg. nitrogen were added, thoroughly mixed, and the whole placed in a 500-c.c. wide mouth bottle. The water content was made up to optimum (2/3 saturation), the bottle plugged loosely with cotton, and incubated at room temperature 7 days. The ammonia was determined by direct distillation in presence of magnesium oxide. If the ammonia recovered by such means is any indication of the relative ammonifying power of the respective floras, it is evident that the difference in nitrate accumulation cannot be ascribed to a difference in this phenomenon. Further, repeated measurements of the ammonia content of the soils as they came from the field, revealed no appreciable difference. This was true whether such determinations were made by direct distillation or by the aeration method of Potter and Snyder (4). TABLE I THE PRODUCTION OF AMMONIA IN SOIL FROM PLOTS NO. 1 AND NO. 9: MG. NITROGEN RECOVERED AS AMMONIA PER 100 GM. OF SOIL Experiments similar to those described above have been conducted to ascertain the nitrate forming power. Table II contains a description of treatment, etc., with the averages of three such determinations. The soil for these experiments was collected October 7, October 23, and November 7, 1914. Samples were incubated for four weeks, the water lost by evaporation being replaced from time to time. Nitrates were then determined by the phenol-di-sulphonic acid method as modified by Lipman and Sharp (2). The results here presented indicate that in the absence of any addition of nitrogen, either with or without an addition of calcium carbonate, Plot No. 1 exhibits a materially higher nitrate forming power. On the other hand plot No. 9 exhibits a materially higher nitrate forming power when nitrogen as ammonium sulphate was added with calcium carbonate, and when nitrogen in form of cottonseed meal and dried blood was added in absence of calcium carbonate. In all other instances the differences are insignificant. A careful study of the data from which this table was con structed reveals, however, consistent differences in only the first three instances mentioned, namely, when no nitrogen was added either in the (ii-14) presence or absence of calcium carbonate, and when ammonium sulphate was added in the presence of calcium carbonate. In all other comparisons sometimes Plot No. 1 and sometimes Plot No. 9 gave higher results. The data contained in Table II indicate very strongly that, if the nitrate nitrogen produced under the experimental conditions here used gives any measure of the relative nitrate producing power, Plot No. 1 certainly contains as active a flora as No. 9. It is only just to add that in a number of experiments carried out in a similar manner, except that the nitrate content was measured at varying intervals, Plot No. 9 gave usually a somewhat more rapid accumulation during the early stages of incubation. However, such large quantities were in all cases formed in soil from Plot No. 1, as to show conclusively that the low accumulation under field conditions could not be attributed to a potentially weak flora. In addition there is absolutely no evidence to indicate a less active flora in the soil of Plot No. 1 when no nitrogen was added. Further, as pointed out above we have never been able to detect, either under field or laboratory conditions, a greater accumulation of ammonia in the soil of Plot No. 1 than in that of Plot No. 9. TABLE II THE PRODUCTION OF NITRATES IN SOIL FROM PLOTS NO. 1 AND NO. 9, The nitrate forming power in solution, as so vigorously recommended by Löhnis and Green (3), has also been tested. Data secured according to this method, for Sept. 14, 1915, are given below. The figures represent milligrams nitrogen converted into NO,. Soil from Plot No. 1 has also been inoculated with soil from Plot No. 9 to see if the introduction of organisms from No. 9 would have any influence on the accumulation of nitrates. The results secured Sept. 14, 1915 are given in Table III. In both these last two mentioned experiments additional evidence is furnished to show that there is but slight |