the measurements, is, however, some indication of the accuracy with which the individual measurements in the machine can be made, even in the case of a coil of insulated wire.

My coil, then, is faulty on account of this ellipticity. I have not yet calculated the possible error due to this cause; it cannot be very great. How much it is only calculation can settle; and the calculation is not without difficulty. But this, after all, is only a question of my apparatus, and does not affect the general question of the possible accuracy of the method.

A more solid metal frame might, no doubt, be turned true to the accuracy required, and if so turned might then be measured with the requisite accuracy in the Whitworth machine.

A still further improvement may be made by making the coil frame of insulating material, say paraffined marble, and winding on it naked wire instead of silk-covered wire. The radius might then be measured with certainty to something much better than 1 part in 10,000. The use of a coil with a single layer of wire instead of many layers greatly facilitates the determination of the mean radius. It, of course, necessitated the finding of a new formula for the coefficient of mutual induction.

This I obtained by direct integration of the general integral for the case in point, viz. a circle and coaxial helix. The coefficient of mutual induction is given us as a sufficiently converging series involving elliptical or quasi-elliptical integrals.

We may, then, in the result conclude that, though I cannot guarantee my own coefficient of mutual induction as correct to 1 part in 10,000 till I have calculated the effect of the coil ellipticity, it lies well within the resources of our mechanical engineers to make a coil and disc free from uncertainty to this degree of accuracy. And if this be so I am warranted in stating that a resistance can be measured in absolute units to 1 part in 10,000.

Now, in the Order in Council, from which I quoted in the early part of my discourse, it is stated that in the use of the ohm standard the limit of accuracy attainable is one-hundredth part of 1 per cent.— i. e. one part in 10,000. Hence the Government gains nothing in precision to compensate for the risk it has taken in adopting as unit the resistance of a standard coil, which may vary from time to time in consequence of changes in the physical condition of the coil.

There is no valid ground left for adopting as ultimate unit any other unit than the absolute unit itself. We have not in our electrical standard legislation given full credit to the mechanical engineer for what he can do for us. He can make a machine that will measure resistance in absolute units with a precision as great as I might even say greater than-that with which the Government is prepared to guarantee its comparisons. Such a machine ought to be at work in the Board of Trade laboratory, in order that there may be opportunity from time to time, at regular intervals, to measure the Government coil or coils in absolute units. It is necessary that this should be

done if we are to be guarded against the perpetual inconvenience of unknown possible changes in the coil, the resistance of which is now the legal unit.

Such a step would have collateral advantages. It would enable the Board of Trade to certify standards of low electrical resistance. With such a machine standards of from a thousandth to a two-hundredthousandth of an ohm may be measured to nearly the same percentage accuracy. All we have to do in dealing with the very low resistances is to pass a sufficiently large current through them, and shunt the standard coil of the machine. But to enter into details on this point would lead me too far. I must content myself with saying that I believe such a machine is much the best instrument for standardising low electrical resistances, and that accurate standards of low resistance would be of great service both in the laboratory and the workshop.

I have, in conclusion, only to express my obligation to my assistant, Mr. Samuel Harrison, for the great and constant help he has given me in the course of my investigations on this subject.

[J. V. J.]

WEEKLY EVENING MEETING,

Friday, May 31, 1895.

SIR FREDERICK ABEL, Bart. K.C.B. D.C.L. LL.D. F.R.S.
Vice-President, in the Chair.

THE EARL OF ROSSE, K.P. D.C.L. LL.D. F.R.S. M.R.I.

The Radiant Heat from the Moon during the
Progress of an Eclipse.

SOME twenty-two years ago (May 30, 1873) I delivered in this room a lecture, the title of which was 'On the Radiation of Heat from the Moon, the Law of its Absorption by our Atmosphere, and its Variation in amount with her Phases.'

I then showed that I had been able, with the aid of a mirror of 3 feet aperture, not only to detect with certainty the heat radiated from the lunar surface, but also to estimate with tolerable accuracy the changes of its amount through the greater part of the lunar month, to ascertain roughly the average quality of her rays as regards refractive index, and to compare their amount at the time of Full Moon with those of the Sun.

I also showed that we had been able by means of a long series of determinations, continued on different nights, through considerable increases and decreases of altitude, to determine with considerable accuracy what may be called the extinction curve for heat with decreasing altitude; probably as completely as Seidel, at Munich, had deduced that for light from his photometrical observations on

stars.

As the phase-curve descended on each side almost to zero on approaching New Moon, it was clear that little or none of the heat

we were measuring came from the interior of the Moon. It was heat derived directly from the Sun.

Some series of determinations were then made alternately with a sheet of glass interposed between the large speculum (of the telescope) and the apparatus and with the glass removed, and from the fact that from 8 to 17 per cent. of the heat (being greatest towards Full Moon) was transmitted by the glass, while some 90 per cent. of the sun's rays passed through the same sheet of glass, it was, we think, clearly established that the heat, which we had already concluded, as stated above, to be directly derived from the Sun was not reflected sun-heat, but heat absorbed and afterwards emitted by the lunar surface.

It might, perhaps, have been expected that, as is more largely the case on the Earth, the highest temperature of the lunar surface

would occur appreciably later than the time of Full Moon, but the phase-curve did not indicate this; on the contrary, whether from some accidental errors or from some real though unexpected cause, it showed a maximum at about 10 hours before Full Moon.

Slide 1 will now be thrown upon the screen, in which the observed phase-curve is compared with Zöllner's curve for light, the latter being much steeper towards the maximum near Full Moon. Zollner's curve agrees more nearly with that calculated for a sphere covered

with meridional corrugations whose sides are inclined 52° from the surface than with the smooth sphere assumed by Lambert and shown in Slide 2. In both cases the surfaces are assumed to be "matt," or free from polish.

In Slide 3 the total heat-curve is compared with that of heat through glass. Slide 4 compares the heat through glass with Zöllner's light-curve, and it will be seen, as might be expected, that

*If we compare these results with analogous cases on the earth, we find that a very much longer time is required before the air temperature, and with it more or less that of the earth's surface, arrives at the final temperature due to the sun's radiation at the time. The maxima and minima of temperature are generally three weeks later respectively than the summer and winter solstices, and the hottest time of the day is generally about two hours after noon. Again, the fall of temperature during an eclipse of the sun, though observations are very con flicting, would appear to be from 7° to 10° only. During the eclipse of July 1878, radiation thermometers gave a depression of 10°, 224o, 32° and 42°, but this is far short of the depression of temperature of the lunar surface, 200°, if the surface has the same radiating power as lamp-black, but probably in reality considerably greater.