and all the Grasses (which form the last quarter of the list) are anemophilous and more or less dichogamous, and therefore not rarely cross-fertilized. Of those which are not anemophilous we notice none which are not habitually visited by insects (except perhaps Gnaphalium uliginosum), and which therefore are almost as likely to be cross-fertilized as close-fertilized; while in not a few (such as the Composite generally and most of the other Gamopetala) the arrangements which favor intercrossing are explicit. There is no cleistogamous and therefore necessarily self-fertilized plant in the list, except Lamium amplexicaule, which also cross-fertilizes freely.

In California the prevalent weeds are largely different from those of the Atlantic States and, as would be expected, are mostly of indigenous species or immigrants from South America; yet the common weeds of the Old World, especially of Southern Europe, are coming in. The well-established and aggressive ones, such as Brassica nigra, Silene Gallica, Erodium cicularium, Malva borealis, Medicago denticulata, Marrubium vulgare and Avena sterilis, were perhaps introduced by way of Western South America. They are mostly plants capable of self-fertilization, but also with adaptations (of dichogamy and otherwise) which must secure occasional crossing.

We cannot avoid the conclusion that self-fertilization is neither the cause, nor a perceptible cause of the prepotency of the European plants which are weeds in North America.

A cursory examination brings us to a similar conclusion as respects the indigenous weeds of the Atlantic States, those herbs which under new conditions, have propagated most abundantly and rapidly, and competed most successfully in the strife for the possession of fields that have taken the place of forest. The most aggressive of these in the Northern States are Epilobium spicatum in the newest clearings, which is dichogamous (proterandrous) to a degree which practically forbids self-fertilization; and in older fields, Asclepias Cornuti, which is specially adapted for cross-fertilization by flying insects; Antennaria plantaginifolia and A. margaritacea, which are dioecious; and next to these perhaps the two wild Strawberries, then Erigeron annuum and E. strigosum, with certain Asters and Golden-rods, all insect-visited and dichogamous, and Verbena hastata, urticifolia, etc., the frequent natural hybridization of which testifies to habitual intercrossing.

Those who suppose that only conspicuous or odorous flowers are visited by flying insects should see how bees throng the small, greenish, and to us odorless blossoms of Ampelopsis or Virginia Creeper and of its Japanese relative.

ART. XXVII.-On a possible cause of variation in the proportion of Oxygen in the Air; by EDW. W. MORLEY, M.D., PH.D., Professor of Chemistry in Western Reserve College, Hudson, Ohio.

PROFESSOR LOOMIS has proposed the theory that certain great and sudden depressions of temperature at the surface of the earth are caused, not by the transfer of cold air from higher to lower latitudes, but by the vertical descent of air from cold elevated parts of the atmosphere. The evidence supporting this theory was published in this Journal in January and July, 1875. It occurred to the writer some time since that if this theory were true, as the evidence makes very probable, the air at the surface of the earth during such a great and sudden depression of temperature might well contain a smaller proportion of oxygen than the average. Dalton, reasoning from the fact that oxygen has a greater specific gravity than nitrogen, argued that the proportion of oxygen to nitrogen in the atmosphere should decrease with increasing altitude above the earth's surface; whether he clearly enough recognized that such a regular decrease would be realized only in an atmosphere in a state of equilibrium undisturbed by convection currents, the writer does not know, not having seen Dalton's memoir. Such a decrease of atmospheric oxygen with increasing altitude has not yet been detected by analysis, although the amount of decrease, on the theory that oxygen and nitrogen are distributed in the atmosphere according to the law which would prevail in case of equilibrium, is so rapid that it would be detected with ease, even in altitudes attained in every holiday ascent of a balloon. This decrease may be calculated from the formula R=R60-9832960 H, where H denotes the height above the earth's surface expressed in kilometers, Ro denotes the ratio of the tension of oxygen to that of nitrogen at the surface of the earth, and R denotes the same ratio for the height H. The constant is computed from the determinations by Regnault of the weights of a litre of oxygen and of nitrogen, and of the specific gravity of mercury. The following table gives in the second column the ratio of the tension of oxygen to that of nitrogen at the height in kilometers given in the first column, and the per cent of oxygen at the same height in the third column. The per centage of oxygen at the earth's surface assumed in the table is that used in the tables for gas analysis in Bunsen's Gasometrische Methoden.

It will be seen that the composition here calculated for a height of a single kilometer is so different from that at the surface that analysis of no very refined accuracy would detect the

variation with ease. But no such variation has been detected even in samples of air collected at the greatest elevations attainable.

But although this is the case, it is certain that in the atmosphere of the same place at different times the oxygen varies by more than one-fortieth of its average amount. Variations so large as this are rare, but variations of the one-hundredth or two-hundredth part are common. It therefore seemed to the writer proper to examine whether facts bear out the conjecture that certain great and sudden local depressions of temperature are caused by the descent of cold air from the upper part of the atmosphere, and that such air may by its poverty in oxygen throw some light on a question in meteorology and a question concerning the physics of a mixture of different gases.

In the number of Wiedemann's Annalen for April of the current year, Jolly has published the results of numerous and very accurate analyses of atmospheric air. He asserts a connection between the variations in composition detected and the direction of the winds when the sample was collected. He considers himself justified in concluding that the atmosphere of tropical regions is poorer in oxygen than that of polar regions, and supposes that at the equator more oxygen is consumed in processes of oxidation than is set free by those of reduction, while the opposite is true near the poles. Since no difference in the composition of the atmosphere at the equator and at the poles has been detected, while on this theory the difference must be one large enough to account for the extreme variations found in temperate regions, and to account for them after such abnormal air had been exposed to admixture with air of a different composition during a passage of thousands of miles, the writer fears that the theory of Jolly will need further proof. Other reasons for a similar doubt will suggest themselves.

On the writer's theory, a sample of air collected at the center of an area covered by a descending current of cold air would

at some given instant be a sample fresh from the upper part of the atmosphere, but little exposed to admixture on the way. If before its descent it had remained at a great height for a long time, it might well have lost some of the oxygen which it contained when it was at some previous time at the level of the sea, and the difference might be enough to be detected. An observer at one side of this central point would have samples more or less mingled with surface air; but even then, a deficiency of oxygen might be detected by accurate analysis. The writer hopes to make arrangements for the regular collection of samples at points which Professor Loomis has indicated as regions of frequent descent of cold air from great heights. But while laying plans for the work, he has thought best to ascertain whether some light on the changes in the constitution of the atmosphere might be obtained by analyzing samples of air collected at home. Having an apparatus for gas analysis lately constructed for the study of the gas issuing from the numerous gas wells of his vicinity, he used this for such determinations. In general, the apparatus is constructed on the plan of McLeod's modification of Frankland and Ward's apparatus. But some important modifications have been introduced, and excellent workmanship was bestowed on details. Some such points but slightly concern analyses by explosion. The connection between the eudiometer and absorption tubes is novel, and has worked well. This will be described in some proper connection. Here will be described everything necessary for a judgment of the accuracy of the analyses to be cited.

The eudiometer and pressure tubes were made from the writer's drawings by Geissler, whose recent death is a loss to science, and a personal loss to so many who have been aided by him. The stop-stocks at the top of these tubes will retain a Torricellian vacuum for weeks. The internal diameters of the tubes are 20·9 and 10.7 millimeters. At the lower end of the eudiometer tube is a glass stop-cock, the use of which is simply to permit the ready cleaning of the tube. Its glass plug is withdrawn, and in place of it is put one of vulcanite so bored that acids or water can be aspirated through the eudiometer without dismounting it and without drawing off the mercury from the pressure tube. The stop-cock at the top of the eudiometer tube is also provided with a similar plug for the same purpose. The pressure and eudiometer tubes are surrounded by a stream of water entering at the top of the pressure tube and running away from near the bottom of the two. The level of water is kept constant by a device similar to that of Thomas, described in a late number of the Journal of the Chemical Society, but perhaps superior to his in some respects. The flow of mercury to and from the movable

reservoir is controlled by an iron stop-cock which is attached to the iron tripod support of the whole apparatus. The plug of this stop-cock is vertical, and is prolonged by a shaft which puts it within easy reach of the observer. By means of a long handle on this shaft, the stop cock can be moved with the greatest delicacy. From this stop-cock, an iron tube, cast in the same piece, extends under the ends of the glass tubes of the apparatus, and two small iron tubes rise from this horizontal tube; these last meet the glass tubes and are connected with them by short tubes of patent black rubber containing no free sulphur. The connectors are tied so as to endure the pressure of mercury having a head of several feet, and are surrounded with mercury so as to be absolutely air tight. The plug of the iron stop-cock is also so surrounded with mercury that the entrance of air is absolutely impossible, and the same precaution was taken at the junction of the two small iron tubes with the horizontal tube. The cast iron of this tube and stop-cock is well japanned, and no leakage through its pores has occurred.

The measurement of the volume of gas in such an apparatus demands an accurate adjustment of the level of the mercury in the eudiometer tube to one of the marks of the graduations. Such an adjustment can be accurately made by admitting mercury very slowly from the reservoir, and closing the stopcock at the required moment. But if now the temperature of the gas be not quite constant, the adjustment can be renewed. for a second reading only by letting in or out a column of mercury of several millimeters, again permitting it very slowly to approach the proper level, and stopping at the instant of contact. It is quite impossible to open the stop-cock, admit the twentieth or fortieth of a millimeter of mercury, note the right instant, and then again close the stop-cock. But for accurate work, the means of altering the level of the mercury by such small quantities, and of doing it by a continuous movement, seemed important. In the end, therefore, of the horizontal iron tube, there works a plunger, packed with great care, which can be moved in or out by a screw. By means of this micrometric movement, the level of the mercury which has been adjusted by the stop-cock can be altered with the greatest delicacy, and readjusted till perfect steadiness is attained. Danger of leakage through the washers around the plunger was prevented by providing a seat into which, when the plunger is screwed quite home, it fits so as to cut off connection between the pressure tube and the rest of the apparatus. The mercury in the pressure tube is, by the use of this valve, always kept at such a height that any possible leakage is that of mercury outward, and not of air inward.