turned in a lathe, and then the thin plates sawn off by means of a small hand-saw in the small wooden support shown in fig. 6. The upper surfaces of the plates may afterwards be rendered perfectly even by a coarse file.

By the aid of these pumice-stone filters many chemical products may be made, the preparation of which has hitherto been almost impossible. For the sake of example I take the preparation of pure dry chromic anhydride; in an hour it is easily possible to filter, wash, and dry crystals of this substance an inch in length. A solution of 2 parts of potassium dichromate in 20 parts of water mixed with 10 parts of concentrated sulphuric acid, deposits, after standing about 24 hours, numerous brilliant needles of chromic anhydride. These may be drained from adhering mother-liquor upon the pumice filter by means of the pump, and in a few minutes completely washed by a small quantity of fuming nitric acid free from nitrous acid. A covering of tolerably strong sheet copper provided with two arms, as shown in fig. 5, is then placed round the tube; by hanging lamps upon the arms the tube may be readily heated to about 60° or 80° C.; and by connecting a chloride-of-calcium tube with the upper end of the glass vessel, a current of dry air may be drawn through the apparatus by means of the pump, and thus in a comparatively short time large and brilliant crystals of chromic anhydride, perfectly dry and free from all impurity, may be easily obtained.

A single pump of the above description costs, including the leaden piping, about 8 thalers (24 shillings); and experience has shown that five or six are amply sufficient for a laboratory of fifty or sixty students. The apparatus, as may readily be seen, can be applied in the operation of evaporating in vacuo; if, however, circumstances will not permit of its being adapted to this purpose, then a fall of 10 or 15 feet is sufficient to filter a precipitate according to the above described method, and so far to dry it that it can be immediately ignited in the crucible. It is therefore not absolutely necessary to employ an air-pump in this process of filtration; any apparatus producing a difference of pressure amounting to a quarter of an atmosphere is sufficient. The simple arrangement represented in fig. 7 is very useful, and is frequently employed in my laboratory. It consists of two equal-sized bottles, a and a, of from 2 to 4 litres capacity, each of which is provided near the bottom with a small stopcock designed to regulate the flow of water. Suppose a filled with water and placed upon a shelf as high above the ground as possible and a' placed empty on the floor, and the two stopcocks connected by means of caoutchouc tubing c, then on allowing water to flow down the tube the air in the upper bottle be

comes somewhat rarefied; and in order to employ the consequent difference in pressure (amounting to a column of mercury about 0.2 metre in height) for the purpose of filtration, it is only necessary to connect the mouth of the upper bottle with the tube of the filter-flask. When the water has ceased to flow, the position of the bottle is reversed, when the operation recommences, So small a pressure as 0-2 metre suffices to render the filter and its contents so far dry that they may be immediately withdrawn from the funnel and ignited without any other preliminary desiccation. The following experiment, made with a portion of the same solution of chromium used in the former determinations, will serve to show the saving of time effected by this simple arrangement:

This amount of chromium sesquioxide (0-2435 grm.) differs from the mean of the former experiments (0.2436 grm.) by one-tenth of a milligramme only, and shows that even by a pressure of 0.184 metre the washing is as complete by the single addition of 26 cub. centims. of water. The duration of the filtering process in the former experiments ranged from 12 to 14 minutes under a difference of pressure amounting to from 0.53 to 0.572 metre; in the last experiment it required 25 minutes under a pressure of 0.184 metre, or about double the length of time. The time needed to analyze potassium chromate in the former case was reduced from 14 hours to 32 minutes; by the latter method the reduction would be from 14 hours to 44 minutes.

The employment of the second method is particularly to be recommended to beginners in qualitative analysis. The experimenter needs only a single funnel, he is obliged to work carefully and cleanly, and the great saving of time and work amply compensates for the little trouble needed to reverse from time to time the position of the bottles.

I believe that the above-described water air-pump will soon become an indispensable piece of apparatus in chemical laboratories. It not only serves as the most convenient method of Phil. Mag. S. 4. Vol. 37. No. 246. Jan. 1869.

C

producing the differences in pressure required to accelerate the process of filtration, and of obtaining the necessary vacuum for evaporation; it is equally adapted to purposes to which neither the mercury nor the ordinary pumps are in any way applicable. By its aid it is possible to calibrate a thermometer with the greatest accuracy, and to estimate the vapour-tension of such corrosive bodies as bromine, chromyl dichloride, &c. by the simplest method possible, in which the necessary operations require scarcely more time than an ordinary determination of a boiling-point.

I purpose returning to these applications of the instrument in a future communication.

II. On a New Form of Permanent Magnet. By FREDERICK A. PAGET, C.E.; M. Soc. Civil Engineers of France; Corr. Mem. Franklin Institute; M. late Government Commission on Chaincable- and Anchor-proving Establishments*.

WT

ITHOUT any distinctly given reason, it is taken for granted in all works on magnetism, and in all the practical applications of magnetism, that it is impossible to magnetize a plate except in the direction of its greatest length.

Michell, in his Treatise on Artificial Magnets,' gives a determinate proportion, but without stating any reason, between the length and the weight of magnets. A magnet, for instance, 2 inches long should weigh one-tenth of a pound. Cavallo recommends a width of one-tenth, Fuss one-sixth, Musschebroeck and, later, Coulomb one twenty-fourth, of the length. In all these cases it is assumed that the direction of the poles must be parallel with the longest dimensions of the solid bar or plate to be magnetized, and that it is impossible to regularly magnetize a square plate, and still less an oblong plate, in a direction transverse to its major axis. That this is correct with a solid continuous plate can be easily proved by experiment; and it is well known to instrument-makers that it is impossible to permanently magnetize a square steel plate. No doubt such results would greatly vary with the constitution and state of the steel employed, the relations of its different dimensions, the mode of magnetization adopted; but the only experiment bearing on the question that I can discover, after much research in scientific works, is that of De la Borne†, who found, on magnetizing steel disks, that as long as they were whole they showed no polarity, and that their polarity only appeared when they were cut in two. Dr. Lamont, in a paper which first appeared in Poggendorff's Annalen (vol. cxiii.), and was communicated to the Philosophical * Communicated by the Author. + Pogg. Ann. vol. Ixxii. p. 26.

this way.

Magazine for November 1861 by the Astronomer Royal, investigated the question of "the most advantageous form of magnets," or that form in which "are united the greatest possible magnetic moment with the smallest possible mass and the smallest possible moment of inertia." In all the forms he experimented upon, the breadth was always less than one-third of the length, and generally about one-fifth; and he does not seem to contemplate the possibility of magnetizing a square plate, and still less an oblong plate, in a direction transverse to its greatest length. Now I find that, by cutting slits nearly up to the middle of a steel plate, a square plate in one piece can with such slits be regularly magnetized; and by this means even an oblong square plate can be regularly magnetized, and with as many poles as may be required, in a direction transverse to its greatest length. I herewith beg to forward a square plate magnetized in It is of watch-spring steel, 0.0075 inch thick and inch inch; it has four pairs of slits inch wide cut from its edges, and leaving a central web inch wide, uniting the whole. On moving a small needle round this square plate, it is seen to be regularly magnetized; and on sprinkling iron filings on the magnet covered by a sheet of paper, they arrange themselves in lines, proving that the magnet really consists of a number of small regular similar magnets arranged below each other in the same vertical plane. On suspending an oblong magnet of this kind with its longer axis in the vertical plane, the needles set themselves to the magnetic meridian; on suspending it flatwise, with its longer axis in the horizontal plane, the longer axis points east and west. As well as can be judged by subjecting them to slight shocks, the magnets are as permanently magnetized as if they were separate from each other. Only time can prove whether they will lose their magnetism. The important question as to what form is the best for retaining magnetism for a length of time is one which, as Dr. Lamont remarks, no one has yet investigated.

Though I have not yet been enabled to try to magnetize a parallelopipedon of steel after slotting it vertically and transversely into a number of bars held together by a central web, I feel very confident that this could be done. The slots could be cut into the parallelopipedon or cube while in a soft state by a thin tool worked to and fro in an ordinary engineers' slotting or shaping machine, and the whole magnetized in a powerful electric spiral, in the way described by Elias. Besides moving the spiral to and fro, as described by him, no doubt in order to overcome the resistance to induction, the cube while in the spiral could also be struck, in order to produce that mechanical vibration which is so favourable to magnetization and demagnetization.

By pointing the needles, or giving them the rhomboidal form, it is evident that, in spite of the poles being nearer (as is well ascertained to be the case in needles of that shape), the ratio of the magnetic moment to the moment of inertia is still higher than in the assemblage of oblong needles. Though, for various reasons, the rhomboid set on its edge is scarcely ever used in practice, an easy calculation shows that it is the most perfect form for a moveable magnet. It seems only to have been tried when lying flat.

Seymour Chambers, Adelphi, W.C.
London, November 20, 1868.

III. Action of Dehydrating Agents on Organic Bodies.
By E. THEOPHRON CHAPMAN and MILES H. SMITH*.

NANE-SUGAR, as is well known, yields carbon on treatment

CANE-SUGAR,

with strong sulphuric acid, which removes the elements of water. In like manner strong sulphuric acid dehydrates common alcohol, yielding olefiant gas. Chloride of zinc removes water from amylic alcohol, yielding amylene. Anhydrous phosphoric acid converts acetate of ammonia into acetamide, and finally into cyanide of methyle. All these are well-known and characteristic examples of the violent dehydration of organic substances. The three short notices which follow describe fresh cases of the action of dehydrating agents.

The first is the instance of nitrate of amyle with phosphoric acid there is very violent dehydration, and pyridine is produced,

C5 H11 NO3-3H2O=C5 H5 N.

The second is the dehydration of formiate of amylamine by means of chloride of zine: there is produced iso-cyanide of amyle, CH2 O2 N C5 H13-2 H2O=CN C5 H11.

The third is a case in which a strong dehydrating agent refused to perform a violent dehydration, viz. the action of chloride of zinc on oxalate of amyle. The equation

was not realized.

C2 04 (C5 H11)2-4H2O=C12 H14

That which actually did happen was as follows:-
C2 04 (C5 H11)2-H2O=C2 03+2(C5 H10).

I. On the Artificial Production of Pyridine.

Pyridine has been produced by Perkin from azo-dinaphthyldiamine (C20 H15 N3) by the action of nascent hydrogen. This

* Communicated by the Authors.