It is seen that, in the case of potassium chloride, the four separate values by the freezing-point method agree well with one another, and the averages of them are almost identical with the dissociation-values calculated from Kohlrausch's conductivity measurements. In the case of sodium chloride, however, the agreement between the freezing-point results of the separate observers is not so complete, and their averages are about four per cent. higher than the values to which the conductivity leads. Whether this latter result arises from a constant error in the freezing-point determinations or from a theoretical inaccuracy in one of the two methods, can not, of course, be determined at present. In view of this fact it must be admitted, to be sure, that these freezing-point measurements do not contribute much towards answering the question regarding the reliability of the conductivity method. Nevertheless a certain significance is to be attributed to the close agreement in the case of the potassium chloride, the best investigated salt.

In another way, however, more decisive evidence of the reliability of that method has been furnished. From the phenomenon of solubility effect it is possible to determine dissociation-values, and experiments of Noyes and Abbot' have shown that the values so determined in the case of certain thallium salts are in accordance with those derived from the conductivity. For details reference is made to the original article; but the final results may be presented here. The first two columns of figures in the following table show the dissociation-values which were calculated from independent solubility experiments with two pairs of salts. The last column contains the corresponding dissociation-values, which were derived from the conductivity. 1 Ztschr. phys. Chem., 16, 136.

These entirely distinct methods, freezing-point-lowering and solubility-effect, lead therefore to nearly the same dissociationvalues. That the electrical conductivity furnishes essentially correct dissociation-values in the case of di-ionic salts in moderately dilute solution, is therefore probable. At any rate, more weighty reasons than those brought forward by van Laar must be discovered before this method should be discredited.

THE ERROR IN CARBON DETERMINATIONS MADE WITH THE USE OF WEIGHED POTASH BULBS.

ΤΗ

BY GEORGE AUCHY.

Received June 2, 1898.

HE difficulty, or impossibility even, of obtaining good carbon results in damp weather has been so often spoken of that anything further on the same subject would seem superfluous. Nevertheless the writer ventures a few remarks. In determining carbon by the dry combustion method it was found that, after the prolong had been in use a few times, a loss of moisture occurred as shown by the gain in weight of a potash bulb containing strong sulphuric acid, placed after the prolong. But it was later discovered that this loss was mainly due to the fact of the copper oxide in the preheating furnace being too fine and packed in too tightly. With coarser copper oxide, and a consequently easier passage of gas and air, and therefore less force and speed in aspiration, the amount of moisture lost from the prolong was much lessened, and the increase in the loss very steady and gradual so that a determination of it by means of a sulphuric acid bulb' would only be necessary once in a while, and might be altogether dispensed with by a frequent change in the calcium chloride of the prolong. This agrees with Dr. Drown's experience who found by many trials that the prolong retained all the moisture passing into it from the potash bulbs; but, while still using the sulphuric acid bulb, some gains in its

1 Or better, a small calcium-chloride tube, as it exposes less surface for the condensation of moisture on a damp day.

weight far in excess of that due to the moisture escaping from the prolong occurred. These were followed in one or two determinations afterward by losses of weight sufficiently large to make the weight of the bulb a minus quantity. This gain in weight could only be attributed to an unusual condensation of moisture on the outside surface of the bulbs, and the loss in weight to a return to normal conditions in this respect. The idea was suggested that the sulphuric bulbs might be made useful in measuring and determining the amount of moisture condensing on the potash bulbs; that is, the amount of moisture condensing on the sulphuric bulbs, found by subtracting the amount of inside moisture gained from the prolong if any, from the total gain in weight of the bulbs, might be taken to be the amount of moisture also condensing on the potash bulbs, and the necessary correction in the weight of the latter could then easily be made. But C. B. Dudley somewhere refers to a method of overcoming this difficulty of moisture condensation suggested by Andrew A. Blair which seemed much easier; this was to have an empty potash bulb on the opposite pan of the balance when weighing. This method of Blair's was accordingly tried, retaining, however, the sulphuric acid bulbs as a check. A succession of damp days gave abundant opportunity, and it was found that the unusual gains and losses in the weight of the sulphuric-acid bulbs still continued to occur. But the empty potash bulbs were not of the same type as those to be weighed as directed by Blair. The potash bulbs were of the Geissler form and the empty bulbs of the Liebig form, for convenience in weighing, since the latter shape leaves the pan. entirely empty for the weights. As the total surface area of the two bulbs was about the same, the difference in shape was thought to be of no consequence, but from the results it seemed that, either this difference in shape caused a difference in moisture condensation; or the passage of gas, air, and moisture, through the potash and sulphuric bulbs, caused a slight heat which gave a different condensation from that on the empty bulbs; or, the condensation of moisture occurred without rule and uniformity; that is, might occur on the one set of bulbs without occurring also at the same time on the others or not to the same degree and vice versa. Tests made by simply

allowing all three of the bulbs stand instead of passing gas and air through two of them as in a determination seemed to show the latter of these theories to be correct. The condensation seems at times to take place entirely at random, so to speak. A condensation on the one bulb does not necessarily mean a condensation on the others, and neither the empty bulbs on the opposite pan nor the weighed sulphuric-acid bulbs are therefore to be depended upon as a means of counteracting, or estimating the moisture condensing on the potash bulbs. The results were as follows:

In taking these weights the empty bulbs on the opposite pan of the balance were used.

Provided the sulphuric bulbs be used to estimate the outside condensation of moisture, and empty bulbs on the opposite pan be also used, what is the error from moisture condensation in a determination? To gain some idea the following tests were made. The first column of results gives the results as they were obtained without any correction for moisture condensation and the very great source of error brought into the process by the use of these sulphuric bulbs, if their gain in weight be always attributed solely to inside moisture from the prolong, is seen by the results of this column. The second column gives the results after being corrected for moisture condensation by the indication of the sulphuric-acid bulbs, assuming the moisture condensing on the potash bulbs to be just equal to that on the sulphuric bulbs. The empty bulbs on the opposite pan were also used.

In the following the "true" result was found by taking the averages of the corrected results:

1 Booth, Garrett, and Blair's result, 1.053. Result by wet method, 1.063. By dry method, 1.05. Average, 1.056.