15h to 17h) is of rather long duration but the amplitude is not excessive, and the principal minimum occurs at about 6 to 8h. The high station of Winnemucca is the only one with pronounced secondary phases. In the north, the characteristics of the curves are similar to those near the center, except that at Ft. Assiniboine and Helena there is a secondary minimum at 21h to 22h and maximum at 24h.

For the Year, in the south, the maxima (culminating at about 15h) are about an average between those for January and July, and a single minimum occurs at 6 or 8h, except for Phoenix which, as in July, has an absolute minimum in the afternoon : the minimum period is rather level. Near the center, but a single moderately sized maximum occurs at about 15 and the minimum is at 6 to 8h. Winnemucca has two maxima, a slight secondary one near midnight; and the morning minimum, although it remains the primary, is not much different from the secondary. At the north, the maxima are usually fairly well marked (crest at about 15") and cover a long period with no secondary phases except at Ft. Assiniboine.

The curves (p. 442) showing the daily march of the wind velocities for January and July for characteristic stations of twenty regions in the U. S. are numbered from 1 to 20 and the following list will identify stations by name:

At 0h. Max. Min. average. At 0h. Max. Min. aver.

In the little table are given for January and July the wind in miles per hour at Oh (midnight), at the time of maximum and minimum wind, and for the average of all the hours of the day.

These curves begin and end at 0h (midnight) and are all in the same scale. At the top of the diagram giving the curves the two hour spaces are indicated by short lines, but the miles per hour have been indicated only in general by vertical scales January.

July. 724

24

at the sides of the curves because it would make the reproduction of the diagram less simple to place the scales properly opposite each curve. The miles per hour of wind at 0h is given at the beginning of each curve.

ART. XXXVI.—The Preparation of Perchloric Acid and its Application to the Determination of Potassium; by D. ALBERT KREIDER.

[Contributions from the Kent Chemical Laboratory of Yale College.-XXXVIII.]

VARIOUS methods for the preparation of perchloric acid have been developed through the long felt want of a process in which the elements of time and danger would be reduced to a minimum and the product increased to quantities commensurate with the growing use of the acid in analytical chemistry. Most of these methods have been found impracticable because of the incidental formation of the dangerously explosive oxides of chlorine, or the time required in refining the product from the impurities introduced with the reagents employed.

Doubtless the best process thus far offered is that of Caspari,* which, however, is to an objectionable degree exacting of time and labor. The product has to be treated and retreated for the removal of potassium and then for the extraction of the hydro-fluo-silicic-acid and at several stages is for this purpose to be left standing for from twenty-four to forty-eight hours. Under the most favorable circumstances it could not be prepared in less than five or six days, and during a great many hours of that time it requires close attention.

The great difficulty has always been with the necessity of a perfect separation of potassium from the perchloric acid, which has been prepared by the ignition of the potassium chlorate. If, for the manufacture of the perchlorate, the chlorate of sodium-which, if not upon the shelves of every laboratory, is nevertheless in the market, almost, if not entirely free of potassium-be used instead of the potassium salt, the complete removal of the base will be unessential; since its presence in the determination of potassium will exert no influence other than that which is beneficial. It is well known that because of its deliquescence and the almost equal solubility of sodium perchlorate with that of the chloride, its separation from the latter by recrystallization from an aqueous solution, as in the case of potassium, is impossible. But the insolubility of the chloride of sodium in strong hydrochloric acid, with the aid of the acid-proof Gooch crucible, affords a means for the liberation of the perchloric acid and the removal of the greater part of the sodium in one operation. Upon this basis, therefore, the following simple method was elaborated.

A convenient quantity of sodium chlorate, from 100 to 300 grms., is melted in a glass retort or round-bottomed flask and

*Zeitschr. für Ang. Chem., 1893, p. 68.

gradually raised to a temperature at which oxygen is freely, but not too rapidly evolved, and kept at this temperature till the fused mass thickens throughout, which indicates the complete conversion of the chlorate to the chloride and perchlorate, and requires between one and one-half to two hours or the retort may be connected with a gasometer and the end of the reaction determined by the volume of oxygen expelled, according to the equation

2NaClO3=NaCl + NaClO 4 +0 2.

The product thus obtained is washed from the retort to a capacious evaporating dish where it is treated with sufficient hydrochloric acid to effect the complete reduction of the residual chlorate, which, if the ignition has been carefully conducted with well distributed heat, will be present in but small amount. It is then evaporated to dryness on the steam bath, or more quickly over a direct flame, and with but little attention until a point near to dryness has been reached, when stirring will be found of great advantage in facilitating the volatilization of the remaining liquid and in breaking up the mass of salt. Otherwise the perchlorate seems to solidify with a certain amount of water, and removal from the dish, without moistening and reheating, is impossible.

After triturating the residue, easily accomplished in a porcelain mortar, an excess of the strongest hydrochloric acid is added to the dry salt, preferably in a tall beaker where there is less surface for the escape of hydrochloric acid and from which the acid can be decanted without disturbing the precipitated chlorid. If the salt has been reduced to a very fine powder, by stirring energetically for a minute, the hydrochloric acid will set free the perchloric acid and precipitate the sodium as chloride, which in a few minutes settles, leaving a clear solution of the perchloric acid with the excess of hydrochloric acid. The clear supernatant liquid is then decanted upon a Gooch filter, through which it may be rapidly drawn with the aid of suction, and the residue retreated with the strongest hydrochloric acid, settled, and again decanted, the salt being finally brought upon the filter where it is washed with a little strong hydrochloric acid. A large platinum cone will be found more convenient than the crucible, because of its greater capacity and filtering surface. When the filter will not hold all the sodium chloride, the latter after being washed may be removed by water or by mechanical means, with precautions not to disturb the felt, which is then ready for the remainder. Of course, if water is used, the felt had better be washed with a little strong hydrochloric acid before receiving another portion of the salt. This residue will be found to contain only an

inconsiderable amount of perchlorate, when tested by first heating to expel the free acid and then treating the dry and powdered residue with 97 per cent alcohol, which dissolves the perchlorate of sodium but has little soluble effect on the chloride.

The filtrate, containing the perchloric acid with the excess of hydrochloric acid and the small per cent of sodium chloride which is soluble in the latter, is then evaporated over the steam bath till all hydrochloric acid is expelled and the heavy white fumes of perchloric acid appear, when it is ready for use in potassium determinations. Evidently the acid will not be chemically pure because the sodium chloride is not absolutely insoluble in hydrochloric acid; but a portion tested with silver nitrate will prove that the sodium, together with any other bases which may have gone through the filter, has been completely converted into perchlorate, and unless the original chlorate contained some potassium or on evaporation the acid was exposed to the fumes of ammonia, the residue of the evaporation of a portion is easily and completely soluble in 97 per cent alcohol and its presence is therefore unobjectionable. One cubic centimeter of the acid thus obtained gave on evaporation a residue of only 0.036 grin., which was completely soluble in 97 per cent alcohol.

Caspari's acid under similar treatment gave a residue in one case of 0.024 grms. and in another 0.047 grms. If, however, a portion of pure acid be required, it may be obtained by distilling this product under diminished pressure and, as Caspari has shown, without great loss providing the heat is regulated according to the fumes in the distilling flask.

Some modification of the above treatment will be found necessary in case the sodium chlorate contains any potassium as an impurity, or if the latter has been introduced from the vessel in which the fusion was made. Under these circumstances the hydrochloric acid would not suffice for the removal of potassium, since a trace might also go over with the sodium and thus on evaporation a residue insoluble in 97 per cent alcohol be obtained. To avoid this difficulty, the mixture of sodium perchlorate and chloride, after being treated with hydrochloric acid for the reduction of the residual chlorate, being reduced to a fine powder, was well digested with 97 per cent alcohol, which dissolves the sodium perchlorate but leaves the chloride as well as any potassium salt insoluble. By giving the alcohol time to become saturated, which was facilitated by stirring, it was found on filtering and evaporating that an average of about 0-2 grm. of sodium perchlorate was obtained for every cubic centimeter of alcohol and that the product thus obtained was comparatively free of chlorides, until the perchlorate was