98 REPORTS ON THE STATE OF SCIENCE.—1917. 
to be CaO, SiO,, 2:5H,O and 4(a0, Al,0;, 12H,O, and microscopical 
examination showed that all the products formed more or less spherulitic 
groups of crystals when the hydration took place in presence of an excess 
of water. Le Chatelier’s explanation was generally accepted. The 
alternative hypothesis, that ‘the calcareous hydraulic cements owe 
their hardening mainly to the formation of colloidal calcium hydrosilicate,’ 
was proposed by W. Michaélis in 1893,? but attracted little attention until 
much later, when it was expanded into a detailed memoir,® at a time 
when the interest in colloidal substances had become much more general. 
According to the hypothesis proposed by Michaélis, the first effect 
of the action of water on the ground cement is the hydrolysis of aluminates. 
An unstable solution is rapidly formed, from which calcium sulpho- 
aluminate, 3CaO, Al,O3, 3CaSO,, H,O (due to the action of the calcium 
sulphate which is present in all commercial cements), and calcium aluminate 
separate as stable phases. The former compound is soluble, and crystal- 
lises readily in needles. The aluminate is less soluble, but is also crystalline 
under normal conditions, forming hexagonal plates. At this stage the 
silicates of the clinker are scarcely attacked. When the hydrolysis 
of the di- and tri-calcium silicates sets in, the only stable products are 
the hydrated metasilicate and calcium hydroxide, of which the latter 
crystallises in large plates, as observed by Le Chatelier. The case of 
the very insoluble metasilicate is different. The investigations of von 
Weimarn‘* have shown that a solid phase separating slowly from a 
moderately supersaturated solution forms well-defined crystals, that 
separation from a highly supersaturated solution favours the formation 
of crystal skeletons or spherulitic groups of needles, and that the 
formation of a highly insoluble product in a_ strongly supersaturated 
solution gives rise to the formation of gels. Michaélis found that the 
hydrated, calcium metasilicate almost invariably took the form of a gel, 
and that even the aluminate assumed that form if the solution were 
sufficiently supersaturated. The coating of the particles of ground clinker 
with a gelatinous sheath is readily observed under the microscope. The 
particles increase in size by the swelling of the gel—a fact which is very 
obvious in photographs taken at different intervals after the addition 
of water.® 
This hypothesis has been challenged on the ground that if swelling 
actually took place, as in the absorption of water by starch grains or 
gelatin, the volume of the cement during setting must increase, and 
that no such increase does in fact occur.6 This objection is based 
on a misunderstanding.’ The individual particles increase in size, but 
1 Thése de Doctorat, 1887, ‘ Recherches expérimentales sur la Constitution des 
Mortiers Hydrauliques,’ Paris, 1904. 
2 Chem. Zeit., 1893, 17, 982. 
3 ¢ Der Erhartungsprozess der kalkhaltigen hydraulischen Bindemittel,’ Dresden, 
1909. Partly in Kolloid Zeitschr. 1909, 5, 9. An account of the hypothesis is given 
by C. H. Desch, The Chemistry and Testing of Cement, London, Arnold, 1911. 
4 The papers of P. P. von Weimarn in the Kolloid Zeitschr. for 1908-9 were collected 
in book form, Zur Lehre von den Zustinden der Materie, 2 vols., Dresden, 1914. 
5 H. Ambronn, Jonind. Zeit. 1909, 338, 270. 
6 ©. Schumann, Yonind. Zeit., 1909, 38, 465. See also A. Martens, Witt. k. 
Material-Priif. Amt., 1897, 15, 109. 
7 A. G. Larsson, Z'onind. Zeit., 1909, 38, 785; W. Michaélis, cbid., 615. See also 
H, Kiihl, ibid., 556, +H 
