2^18 Moiis^ System oj Lrystailography 
cases equal, we divide each of them into ten equal parts ; one 
soon learns to value those parts with tolerable accuracy, and 
each of them is expressed by a decimal fraction, whenever the 
hardness is to be designated in numbers. Thus the hardness 
of a mineral, midway in that respect between felspar and 
quartz, will be expressed by 6.5, and so for others. 
68. Employment of the Scale, — When the hardness of a mi« 
neral is to be determined, we first of all select from the scale 
those two numbers between which it stands, in order to com- 
pare it with them. This may be done by trying the scale, from 
the top downwards, till we hit upon some member of it 
which may be scratched by the given mineral, but which does 
not scratch it. We then take that member, the next above 
it, and the given mineral ; and, having chosen from each a 
fragment pretty similar in form, and, so far as possible, simi- 
lar in structure to the other two, we rub all the fragments 
oil a very delicate Jik, The smallest possible force must be 
applied in this operation ; the sound and feel will enable us 
to judge at what distance, in respect of hardness, the given 
mineral stands from both the members of the scale. This ex- 
periment may be repeated and altered, till the result can be 
regarded as certain, and depended on for any use to be made 
of it afterw’ards. The hardness of the file is so far indifferent, 
as it is not with that instrument that the mineral is compared, 
but by means of ii^ with the other members of the scale. For 
the higher degrees, however, the file ought to be as hard as 
possible. Practice will convince any one of the sufficiency of 
this method for the natural history ^ the mineral Jdngdom, 
and the experience, on which it ojrtainly rests, may be acquir- 
ed very soon. It resembles an imperfect instrument, which 
gains in correctness by the dexterity of him who applies it. 
69. Lmtre and Colour, — .It has struck us, as necessary to 
settle the varieties of lustre in a manner somewhat different 
from their previous mode of arrangement. We divide metallic 
lustre into perfect and imperfect. The first is found in na- 
tive metals, the second in uran-ore, tan tali te, atid others. In 
characterising orders, the terms metallic^^ and 7ion-meiaL 
lief are introduced. The former includes only the complete 
