RATIO OF ARGENTIC TO STRONTIC BROMIDE.

In many of the preceding determinations the bromide of silver resulting from the decomposition was weighed. In every case a slight excess of silver nitrate was added, to render the argentic bromide wholly insoluble in the filtrate. The very slight amount which may have been dissolved by the wash water during its brief contact with the precipitate was not considered. The precipitate was collected upon a Gooch crucible; and the traces (0.04 to 0.2 milligram) of asbestos carried through were collected upon a small washed filter, ignited separately, weighed, and added to the gain in weight of the crucible. From this was subtracted the loss in weight of the precipitate upon fusion in a covered porcelain crucible. A description of the dark room used for the experiments, and many other precautions and details, will be found in other papers." The results are tabulated below.

* These Proceedings, XXVIII. 24; XXIX. 74.

It remains only to bring together the results into one table.

Total average

Average, rejecting I. above = 87.663

The last average is probably most nearly correct.

The analysis of strontic chloride has already been begun, and the preliminary results indicate that the results given above are certainly not too high. For the present, then, the atomic weight of strontium may be taken as 87.66 if oxygen is 16.00, 87.44 if oxygen is 15.96, and 87.01 if oxygen is 15.88.

XV.

ON THE ELECTRICAL RESISTANCES OF CERTAIN POOR CONDUCTORS.

BY B. O. PEIRCE.

Presented October 10, 1894.

SINCE the subject of electricity began to be studied seriously, many experimenters have made lists of substances arranged in the order of their electrical conductivities. These lists have not agreed with one another in all respects; but at one end of every one of them metals have stood, and at the other end such insulating substances as ebonite, glass, paraffine, shellac, and mica. Somewhere between these extremes have appeared the so called "half-conductors," * like wood and some kinds of stone. How these latter are to be classed depends very much, of course, upon the uses to which they are put. For work with the small charges and high potentials of experiments in electrostatics, we must generally consider wood as a conductor; while, for practical purposes, we may regard the wooden base upon which a telegraph instrument is mounted as a perfect insulator.

In making electrical measurements in the laboratory, it is often necessary to be able to change quickly the connections of one's apparatus, and for this purpose some kind of "switch-board" must generally be provided. Sometimes a dry wooden board, into which holes have been bored to form mercury cups, will suffice; sometimes a plate of ebonite or a non-combustible slab of slate or marble is required.

I have been obliged, during the last three years, to procure several hundred more or less complicated switch-boards, and many of these had to be used in making accurate measurements of electrical quantities. It has been necessary, therefore, to determine under what circumstances hard dry wood or red vulcanized fibre may safely be used, and when marble or ebonite, or even a block of freshly scraped paraffine, is required. For use with these switch-boards I have provided several hundred resistance coils of German silver, platinoid,

* Du Moncel, Annales de Chimie et de Physique, [5.], X. 1877; Addenbrooke, Muir and Jamieson's Pocket-Book, p. 194.

and manganine wire, wound on spools two inches in diameter over all, and from four to eight inches long. The coils are furnished with stout copper terminals (Figure 1), and are protected by cylindrical shields made of brass or pasteboard tube. The copper terminals are screwed, with axes parallel to each other, and one inch apart, into one end of each spool. Spools and shields together act as shunts to the coils, and it has been necessary to determine a lower limit for the resistance which such a shunt could offer in practice. It has been necessary also to measure the insulation resistance between the two mercury cups, on a switch-board, into which the terminals of one of the resistance coils dip. These cups are formed by holes three eighths of an inch in diameter (Figure 2), and five eighths or three quarters of an inch deep. The axes of each pair of holes are one inch apart. The holes are bored, or drilled, in a lathe, in the top of the switch-board, with a Förstner bit if the material be wood or vulcanized fibre or ebonite, with a flat or twist drill moistened with water if slate or marble be used.

о

FIG. 1.

In order to get such information as I needed with respect to the insulation resistance which might safely be counted on in the case of wood, marble, or vulcanized fibre, used for the purposes just described, I made a long series of measurements, with the help of a battery of twelve dry cells, and an absolutely calibrated high-resistance mirror galvanometer, on a large number of specimens. Of course individual measurements of this sort have little general value, but a large number of experiments on different samples of material of a given kind make it possible to set a lower limit to the resistance of

о

this substance when used in a given way, and such knowledge as this is often useful when one is planning apparatus. FIG. 2. For instance, my experience seems to show that it is perfectly safe to assume that the specific resistance per cubic centimeter of an inch-thick slab of pure white Vermont marble, which has been standing exposed to the air in a fairly dry room for three weeks, is not less than 10° ohms, and is probably as much as 1010 ohms. Really dry white marble has a far higher specific resistance than this. It is safe also to set a lower limit for the specific resistance of sea

soned wood of a given kind, but only in the sense that wood of resistance as high as this can always be obtained without difficulty in the market. Abnormal specimens occur. In measuring the insulation resistances between the copper terminals of unparaffined maple and birch spools, like or similar to the one represented in Figure 1, I experimented upon a large number of spools which had been lying for about a year in a certain dry closet. The smallest resistance in any case was 1,100 megohms, and the average resistance was more than 2,000 megohms. This is what one may expect to get in spools of this kind. In the same closet were some spools of a different lot, bought of the maker of the other spools, and in no way different in appearance from them. These also had been seasoning beside the others for a year, yet the average insulation resistance of these spools was only a little over one megohm. This is an extreme case. The nearest approach to it that I found in experimenting on other lots of spools was that of some which had been standing for a long time in the damp basement of the laboratory, and represent what the ordinarily good dry spool might become if it were placed for months in a moist place. Yet the lowest resistance in the case of these spools was more than 100 megohms.

It is, of course, well known that the insulation resistance of a porous half-conductor depends very much upon the amount of moisture which it contains, and that this moisture may give rise to all manner of anomalies, as Du Moncel has shown. Thus, white marble when it comes from the mill is often a fairly good conductor, owing to the water which it has absorbed in the process of manufacture, but a fortnight's drying in the sun sometimes increases its resistance ten thousand fold. It is now almost always possible to get kiln-dried wood, and after wood or marble has once been thoroughly dried, an immersion in a bath of hot paraffine tends to prevent the reabsorption of moisture. Red vulcanized fibre absorbs hot paraffine greedily; but I do not think that it would be easy to saturate a piece of fibre so thoroughly with paraffine that a drop of water allowed to rest on its surface for a few moments would not begin to raise a blister.

Prolonged immersion in clean, hot, melted paraffine always increases the insulation of a half-conductor, even if the bath leaves no perceptible coating on the outside. This increase, however, is very slight in the cases of some close-grained substances like rosewood, though it may amount to three or four times the original resistance in the case of a porous conductor. If while such a conductor is immersed in hot paraffine the bath and its contents be placed under the receiver