482 



NA TURE 



[September 5, 1907 



in the velocity of the particles alter the distance between | 

 the bright patches in Fig. 2, giving the effect of changes \ 

 of luminosity running round a circle, a well-known auroral 

 phenomenon. Other phenomena analogous to the " danc- 

 ing " of auroral rays are easily produced by advancing or 

 withdrawing a bar magnet. M. Villard seems to regard 

 magnetic disturbances rather as agents causing auroral 

 phenomena than the converse, but his theory seems still in 

 a somewhat uncrystallised state. Those actually familiar 

 with auroral phenomena will think of a number of points 

 requiring explanation to which M. Villard does not refer ; 

 but his researches, like those of Prof. Birkeland, are at 

 ?iU events highly interesting and suggestive. 



C. Chree. 



CHEUISTRY AT THE BRITISH 

 ASSOCIATION. 

 T^HE proceedings of this year's meeting were largely 

 ■*• influenced by the number of foreign visitors, most 

 of whom made lengthy communications to the section. 

 The latter part of the president's address in particular 

 elicited the warmest approval ; it was felt that his warn- 

 ing against the danger we now run into of neglecting to 

 cultivate manipulative skill in the young chemist was 

 most timely. The discussions ranged over an unusually 

 wide field ; that on valency came first in importance ; in 

 its course the new doctrine of Barlow and Pope was 

 subjected to searching criticism, although the general tone 

 was far from hostile. Of deep significance, as illustrating 

 the importance of maintaining the very closest connection 

 between science and industry, was the discussion on the 

 quality " strength " in flour; it was felt by those con- 

 cerned in the debate that this did much towards establish- 

 ing clear ideas on this controversial subject. 



Prof. Pope opened the discussion on valency with an 

 account of the theory put forward by Mr. W. M. Barlow 

 and himself. Combining chemical with crystallographic 

 data, it is possible to show that the fundamental valency 

 of an element represents, very approximately, the number 

 of units of volume contained in the polyhedral cell which 

 constitutes the domain of its atom in any molecular 

 assemblage. Valency, from this point of view, is primarily 

 a simple volume relation. If the atoms are regarded as 

 centres of attraction and repulsion, a crystalline structure 

 may be considered to be an equilibrium arrangement of 

 such centres. If each atom be allotted its own poly- 

 hedral cell or sphere of influence, the simplest assumption 

 that can be made as to the shape of these spaces is that 

 they depart as little as possible from sphericity. The 

 polyhedral cells must fill space without leaving interstices, 

 and the centres of the identical polyhedral cells will be 

 those of the centres of a closest packed assemblage of 

 equal spheres. There are two varieties of homogeneous 

 arrangement of equal spheres in closest packing, dis- 

 tinguished as the cubical and the hexagonal closest packed 

 arrangements. 



A crystalline element, in which no molecular aggrega- 

 tion of the single atoms occurs, should thus exhibit holo- 

 hedral cubic or hexagonal symmetry. Of the forty known 

 crystalline forms of the elements, only six are neither 

 cubic nor hexagonal ; in the rest the axial ratios all 

 approximate to the theoretical values calculated on the 

 above hypothesis. The binary compounds can be con- 

 structed from spheres of two kinds, but of appro.ximately 

 equal size, present in equal numbers. 



This hypothesis has been worked out in detail in the 

 case of the alkali haloid salts and of silver iodide. The 

 assemblages thus constructed present geometrical proper- 

 ties which are entirely in harmony with corresponding 

 physical properties of the crystalline material. The 

 cleavage, twinning, and gliding of the cubic alkali haloid 

 salts and the dimorphism of silver iodide are precisely 

 paralleled by corresponding properties of the assemblages 

 suggested as representing the crystalline structures of these 

 salts. 



Each crystalline substance is to be regarded as a close- 

 packed homogeneous assemblage of the spheres of atomic 

 influence. 



Prof. Sollas criticised the form of close packing put 

 NO. 1975, VOL. 76] 



forward by Messrs. Barlow and Pope, and objected to the 

 structure suggested for silver iodide on the grounds that 

 it did not explain the sudden contraction in volume of 

 this substance when heated to a certain temperature. 

 He described and showed models to illustrate an alter- 

 native system. Prof. Miers agreed that the fact that more 

 than 80 per cent, of the elements and binary compounds 

 crystallise in the cubic system was a strong argument in 

 favour of closest packing, yet this principle must be tested 

 by all the physical properties as well as the geometric 

 form of the crystals. Dr. Tutton also criticised the idea 

 of closest packing ; ho suggested that the topic axial ratios 

 should be taken as an index and measure of the relative 

 closeness of the packing. 



I'rof. Abegg pointed out that, according to van der 

 Waals, in the liquid state three-eighths of the volume are 

 occupied by matter and five-eighths are free space, and that 

 as substances changing from liquid to solid do not vary 

 essentially in volume, this would seem opposed to the 

 idea of close packing. In answer to the above, Mr. 

 Barlow explained that the spheres were only in reality 

 centres of influence and attraction, and thus there was 

 no real distinction between loose and tight packing. 

 Prof. Tilden remarked that the authors in their 

 explanation of what happens when a change of valency 

 occurs did not appear to recognise that there are two 

 kinds of valency, the one corresponding to the electric 

 charges and the other an extra valency developed under 

 special conditions. 



Prof. Larmor thought that the views of the authors 

 seemed to carry weight in their application to the forms 

 of crystals. Tlie approximation that is exhibited by actual 

 crystalline forms to those calculated for dyad molecules 

 consisting each 01 two equal spheres or nearly spherical 

 domains of influence seemed noteworthy ; it gained even 

 more weight when it was remembered that considerable 

 difference in size of the spheres would produce but little 

 difference in the ratios, so that the restriction to equality 

 of the domain of influence could be largely dispensed with 

 and yet the results remain substantially as they are. He 

 hazarded the suggestion that the direct operations of the 

 chemist can grasp only those molecules which have pro- 

 nounced architectural features, but that there may be 

 whole regions of incipient combinations which do not 

 submit to architectural modification, though they may be 

 recognisable indirectly, as by the spectroscope. 



The electrochemical aspect of valency was to associate 

 it with the number of loose and displaceable electrons in 

 the atom. The essential physical features of a metallic 

 atom being generally that a certain number of its negative 

 electrons are thus relatively free, we may imagine that 

 when it is in combination to form a molecule of a salt 

 these electrons are attracted across into the domain of 

 llie radicle thus united with it, and when accidental 

 separation occurs in the appropriate way they may remain 

 there. But while something like this is a fact, the 

 mechanism remains largely a mystery, as it has been 

 ever since valency was first connected in this way with 

 electric displacement by Faraday eighty years ago. 



Prof. Werner (Zurich') and Prof. Abegg (Breslau) gave 

 descriptions of their views on the broader questions of 

 valency. Prof. Kauffmann (Stuttgart) spoke on the 

 divisibility of valency, and dealt briefly with his own 

 investigations on the colour and fluorescence of organic 

 compounds in connection with structure. 



The second morning was entirely given up to a joint 

 discussion with Section G on explosion temperatures, in 

 which Profs. Boudouard, Haber, Smithells, and H. B. 

 Dixon spoke on the chemical side. In the afternoon 

 Prof. Dixon described his observations on the ignition 

 point of various gases and mixtures. In the discussion 

 which followed Mr. Dugald Clerk dwelt on the trouble 

 arising from pre-ignition in large gas engines, and showed 

 that it was important to be able to predict the ignition 

 temperature from the nature of a mixture of gases, and 

 so to allow proper compression space in designing the 

 engine. 



Dr. H. O. Jones gave a concise account of work carried 

 out with Sir James Dewar on iron carbonyls. The 

 pentacarbonyl is a yellow liquid, which resembles nickel 

 tetracarbonyl in its properties, but is more stable towards 



