8 REPORT ON THE COMPLETION OF 
proved, it is evidently contrary to all sound investigation to incur the risk of ills 
we know not of. I feel inclined to attribute some curious discrepancies betwixt 
the results obtained in town and country observatories to some such causes. 
The building to contain the variation instruments and computing room, 
should be made of stone if possible, even though it should contain some small 
quantity of iron. It is of the greatest importance that the temperature within 
the instrument room should be nearly uniform; for this end, stones or logs of 
wood are essential ; I do not believe that any small quantity of iron which ana- 
lysis might detect would affect the variation results. In the construction of the 
observatory, however, an important matter has always been lost sight of: it is 
nearly impossible that the temperature in our climate can be very uniform 
without a complete exclusion of the external atmosphere. Such an exclusion has 
been attempted by sealing up the observatory, and by burying it under ground ; 
methods which might have served for the instruments, but which would in a short 
period have served for the observers too. It appears essential, then, that the in- 
struments should be placed in a different room from the observer: this room 
should be wholly within the observatory, with passages or rooms around it, 
separating it on all sides from the outer wall: the telescopes could be placed in 
the wall of this room, the eyepieces being outside; a single glazed aperture with 
a lamp and proper arrangement of mirrors would serve to illuminate all the 
scales. There should be no windows, and only one (double) door into the room 
which could be well closed, and would require to be opened only at considerable 
intervals. The roof of the observatory should slope only to the north, so that the 
sun could not beat on it, or some arrangement should be made by a double roof, 
to prevent the heating there from atfecting the internal temperature. 
By such arrangements, the diurnal variation of temperature would be 
scarcely appreciable within the magnetometer boxes. It would be necessary, in 
order to determine the temperature co-efficients, either that this room could be 
opened for some time to the variations of the external temperature, or that it 
should be possible to heat it artificially: this might be done by a stove, or by 
pipes with hot water laid in the external room, which could be used as a com- 
puting room. The annual variation of temperature might be very much dimi- 
nished, and the dryness of the instruments be insured by this arrangement. It 
is to the effects of varying temperature and humidity, that the principal errors 
are due for the three variation magnetometers. 
Having suggested what I conceive to be the best position for the variation 
magnetometers, I would remark, shortly, with reference to the dry and wet bulb 
thermometers, that they ought not to be placed in a recess to the north; a posi- 
tion which, although sufficiently shaded from the sun, is, however, in general 
wholly abnormal ; it is generally damp, and if the house be of stone, of a consider- 
