50 REPORT—1850. 
only when the equivalent of magnesia salt does not exceed 6 grains of carbonate of 
lime in a gallon of water. 
2. That the degrees of hardness of an ordinary water cannot be inferred by the 
rule,—‘‘ compute the grains of lime, magnesia, oxides of iron, alumina, ina gallon of 
water, each into its equivalent of chalk. The sum of these equivalents will be the 
hardness of the water.” 
3. That the degrees of hardness of a water containing magnesia and lime salts, as 
shown by the soap-test as it is now applied, cannot, in almost every case, be taken as 
representing the amount of these salts in the water; nor, in nearly every instance, can 
it be considered as giving the amount of lime in a water when magnesia is present. 
4, That water might show by the soap-test a small degree of hardness in compari- 
son to the considerable quantities of salts of magnesia and of lime it might contain ; 
and trusting to this method of analysis alone, when selecting water for ordinary use 
and for steam purposes, might lead to a water being selected which might not be 
conducive to the general health, and which would leave considerable deposit in vessels 
in which it was boiled—a great deterioration to its use in steam generating. 
Remarks on the Isomorphous Relations of Silica and Alumina. 
By Prof. Cuarpman, Univ. Coll., London. 
In this paper the author endeavours first to disprove the hypothesis, adopted by 
certain chemists, that 3 atoms of alumina are isomorphous with 2 atoms of silica. He 
shows, that, on this supposition, the formula RO, R? 0%, which represents the highly- 
natural group of aluminates, ferrites, chromites, and their isomorphs,—comprising the 
well-established monometric or regular forms, spinel, gannite, magnetic iron ore, 
chromic iron, &¢.,—should be isomorphous with 3RO, 2Si0%, the formula of a remark- 
able division of monoclinic or oblique prismatic silicates, including augites, Wollasto- 
nite, and other kindred forms. The minerals of these respective groups, however, do 
not bear the slightest analogy to each other. 
The author then proceeds to examine the opinion which makes alumina isomor- 
phous with silica atom for atom. He shows, that, according to this view, the above 
formula, RO, R?0%, of the spinels, magnetic iron ore, &c., should be equivalent to 
3RO, Si0%+ Al? 0%, 5105, the formula of the garnets. Here the included minerals 
belong equally to the monometric system; and although the spinel group affects 
octahedral forms, whilst the garnets usually crystallize in rhombic dodecahedrons, 
yet certain members of the series offer transitions from one to the other; magnetic 
iron ore, for instance, which occurs both in octahedrons and in rhombie dodeca- 
hedrons. 
Allusion is then made to the statement of Berzelius, that he had analysed speci~ 
mens of magnetic iron ore united crystallographically, as it were, to specimens of 
garnet, the two constituting a single crystal; also to the fact, that the author had 
seen rhombic dodecahedrons of garnet in which each of the three-sided angles—the 
positions of octahedral faces—consisted of magnetic iron ore. These facts, tending 
in a manner to confirm the second hypothesis, are met by the circumstance, that 
unions of a somewhat similar kind between minerals of dissimilar crystallization are 
not unknown, as in the familiar disthene-staurolite crystals from St. Gotthardt; 
although, in this latter case, the two crystals might be referred to the same group, by 
considering monoclinic forms as hemihedrons of the trimetric system, according to 
the views of Professor Weiss. 
The angles of the principal monometric furms, however, being necessarily constant, 
no matter what the nature of the substance may be to which these crystals belong, the 
isomorphism of the two formule cannot be absolutely proved from the spinel and 
garnet minerals. The author therefore relies more particularly, in confirmation of 
this view, upon the crystallographic relations exhibited by two other minerals allied 
by their composition to the above. These minerals are the Hausmannite and the 
Idocrase,—the first having the composition MnO, Mn? Οϑ, and consequently belonging 
to the spinel series; and the second having the general composition of the garnets. 
Both crystallize in the dimetric or square prismatic system, the Hausmannite occurring 
in only three forms, square-based octahedrons. Taking the more frequent of these 
᾿ 
