84°5. These are practically identical with the result of the larger figure I originally made.

Fig. 2.-Showing the shift of the isothermals in the epoch of
great eccentricity, as calculated by the third method.

There is one, and, as far as I can see, only one, objection that can be made to the results I have obtained. It might be urged that the winter heat at present received by lat. 54° from ocean currents is much in excess of that received in the epoch of great eccentricity by lat. 50°, and that therefore lat. 54° at present may be much warmer than lat. 50° was in the earlier epoch. The observations already made on

ocean currents seem a sufficient answer, but happily the objection can be entirely disposed of. For the transference of heat from ocean currents to lat. 54° at present depends on the excess of temperature of the ocean over what I may call the sun-heat temperature of 54°. Hence the objection supposes that the present midwinter excess of ocean temperature at 54° over the sun-heat temperature at 54° is greater than the excess of ocean temperature at 50° over the sun-heat temperature at 50' was in the epoch of great eccentricity. But since the present midwinter sun-heat temperature at 54° is the same as that at 50° in the earlier epoch, this supposition requires the midwinter ocean temperature to be now higher at lat. 54° than it was at lat. 50° in the epoch of great eccentricity. But since the sun-heat temperature of the ocean at lat. 50° at the earlier epoch was the same as that at lat. 54° now, there is nothing which could tend to make the winter temperature of the ocean at 50° in the earlier epoch lower than that at 54° now, while the far greater summer heating the water then underwent in the lower latitudes must have made its winter temperature higher than it is now. Hence the supposition that the winter temperature of the ocean would be lower is disposed of for Great Britain at least *.

Of course the foregoing argument proceeds on the supposition that the configuration of the land, and with it the general character of the ocean currents and air currents was the same at the epoch of great eccentricity as it is now. It then supplies a satisfactory proof that there is nothing in the astronomical causes which would alter those currents, or at least nothing of the nature required by Croll's theory.

If, instead of taking the radiation proportional to the absolute temperature, we take Stefan's law, published in 1881, giving it as proportional to the 4th power of the temperature, we should reduce Croll's result by one fourth. Thus his 45°-3 F. would come down to 11° F.; and if we further consider that ocean and air currents are twice as effective as winter sun-heat in maintaining the temperature we should get a lowering of about 4° F. at midwinter, about the same as the amount I arrive at otherwise. This is not quite accurate, for we ought to take the 4th power from zero, not from -239, and also we ought to take the percentage diminution, not of the solar heat, but of the total heat. Still the calculation may help in a rough way to show the reasonableness of the results obtained in the paper.

* Of course in strict accuracy we should use in the argument not 54° but the somewhat lower latitude whose midwinter sun-heat temperature at present is the same as that of 50° in the period of great eccentricity.

LXIV. Description of a very Sensitive Form of Thomson Galvanometer, and some Methods of Galvanometer Construction. By F. L. O. WADSWORTH*.

[Plate XIV.]

S considerable interest has of late been manifested, particularly in Germany, in the construction of very sensitive galvanometers, a brief description of one, which has recently been constructed for the bolometric work of the Observatory, may not be out of place in connexion with my last article (p. 482). This galvanometer in question is of the Thomson type of construction with four coils wound under my instructions by Messrs. Elliot, Bros.

Fig. 1, Plate XIV., is a section through one of the coils showing the contour of the individual sections, of which there are five, of about 4 ohms each, in each coil. The size of wire, number of turns, and resistance in each section of one of the coils, which is typical of all of them, is given in the following table.

The radius of the inner coil is 2 millim., and that of the outer coil 50 millim., and the depth of the coil was about 40 millim. As will be seen from the above table, the diameter of the wire increases somewhat less rapidly than the mean radius of the section, as required by Maxwell's theory, the thickness of the insulating covering, while not constant, being proportionally thicker for the finer than for the larger sizest. The total number of turns in the four coils is 9593, and the total resistance (in series) 86 ohms at 20° C.

* Communicated by the Author.

† Maxwell, Elec. and Magnetism, vol. ii. p. 363. Phil. Mag. S. 5. Vol. 38. No. 235. Dec. 1894.

2 P

Each coil was cemented by means of melted shellac into an open brass case, which left the windings on the face of the coil exposed, and enabled them to be brought as close to the needle system as possible. These brass cases screwed into L-shaped supports, which rested on three adjustable screwsa, b, c, Plate XIV. (placed at three corners of the foot-plate of the L), the points of which slide in V-grooves planed in the metal plate which forms the base of the galvanometer-case. They are clamped in any desired position by means of a screw, d, working in a slot in the plate, as shown. This means of support allows the coils to be accurately centred with respect to each other, and, in conjunction with the levelling-screws on the case, to the needle system, independently of the adjustments of the latter. It also enables the coils to be readily removed whenever it is necessary to get at the needle, and the distance between the coils to be varied to increase or decrease the delicacy, without altering the astaticism of the system by means of a directing magnet.

The needle system itself is shown in fig. 4 (PI. XIV.). The central staff is about 150 millim. long, drawn from glass tubing, and weighs about 5 mgs. On account of the length and thinness of the staff, special means of drawing it were necessary in order to get a perfectly straight piece.

A good method is to clamp a selected piece of tubing, about 5-10 millim. in diameter, in a retort-stand so it hangs vertical, and attach to the lower end a 4-5 lb. weight, which rests on some simple form of trap 4 or 5 feet above a box filled loosely with waste or shavings. The tube is heated uniformly by two good Bunsen burners until it begins to soften, then the burners are removed, the trap is inmediately sprung, and the weight falls into the box placed to receive it, drawing out a thin tube of glass, the diameter of which will depend on the length of tube which has been softened. From a few fibres thus drawn a piece can be selected which will be satisfactory as regards straightness and lightness.

The two members of the system are built up each of ten small magnets, five on each side of the staff, the central one about 3 millim. long, and the upper and lower ones each a little less than 2 millim. They are made from the smallest size sewing-needles broken to the required length, but otherwise untreated *. They were attached to the staff by first

*This material is unsuited for the purpose, being too soft a grade of steel for retaining a high permanent magnetization. Some bars of special magnet steel were ordered, but have not yet been received, and pending their arrival the above material was used as the best available for the purpose. With steel of proper size, quality, and hardness, the magnetic moment could, I am certain, have been more than doubled, without any increase in weight.

cementing each set of five, in proper position, to a piece of thin tissue-paper, placing two sets face downward in the proper position, and at the proper distance apart, on a glass "flat," laying the glass staff on top of them, and cementing it to each by means of a very small drop of thick shellac. When dry enough to handle, the other two sets were attached to the other side of the staff, opposite the first two, in a similar manner. The mirror was then attached midway between the two members by means of a minute fragment of soft wax (" universal "), which touched the upper edge only of the mirror; a method more satisfactory than any other I have tried for mounting small thin mirrors without distortion.

This mirror was 23 millim. in diameter and 1 millim. radius of curvature. It had an accurately worked surface (by Brashear) (rendered necessary by the fact that it is used for a photographic record), and had to be therefore quite thick and heavy for its size. The weight of the mirror was about 12 mgs., and the weight of the whole system about 40 mgs. It was magnetized and astaticized after being completed by the method described in my last article. The system was suspended by means of a fine quartz fibre, about 40 centim. long, whose tension was negligible. The method of supporting the fibre is in some respects novel, and has proved very successful in eliminating vibration, which, on account of the lightness of the system, at first proved very troublesome. The glass tube, f, of about 1 centim. bore, which carries at its upper end the adjustable head, h, to which the fibre is directly attached, is not directly connected with the galvanometer-case, but is supported by two thick rings of soft rubber, m, n, very slightly compressed between the glass tube and an outer heavy brass tube which is screwed to the top of the galvanometer-case. There is consequently no metallic or solid connexion between the fibresupport and the rest of the instrument, and the vibration which is communicated to the latter from the pier is absorbed by the rubber before it reaches the needle. It is also possible with this arrangement to attach the directing magnet to the outer brass tube without prejudicing the steadiness of the image during adjustment.

The damping of the needle is effected partly by a piece of dragonfly's wing attached to the back of the mirror and partly by four copper rods, which slide into the cores of the coils.

The coils on each side are connected in series, and the terminals brought up and connected to two copper bindingposts on the top of the case. The two sides could therefore