July :i, 1910J 



NATURE 



71 



The Sterilisation of Liquids by Light of very short 

 Wave-length. 



DiRiM'. ihe past ycai- several articles have appeared in 

 the CoDiptcs rendus Jes Seances de VAcademie des Sciences, 

 Paris, on tlie sterilisation of liquids by ultra-violet light. 

 The notes of M. Billon-Daguerre have particularly attracted 

 my attention, since he has endeavoured to utilise the region 

 of the spectrum discovered by Schumann for the sterilisa- 

 tion of water. It is obvious that the question of the 

 transparency of water for light of very short wave-length 

 is important in this connection, and, as there seems to be 

 no data which bears on the matter, I have recently made 

 some experiments. 



I used a vacuum grating spectroscope arranged in the 

 same way as when I investigated the transparency of some 

 solid substances. The water was distilled, but without 

 any special precautions, and was enclosed in a cell with 

 fluorite windows. Two of these cells were employed, one 

 giving a water column of half a millimetre, the other giving 

 a millimetre column. With the half-millimetre cell in the 

 light path the spectrum was cut off at \ 1792 (Angstrom 

 units), even after a prolonged exposure. It appeared that 

 this limit of the spectrum receded rather slowly toward the 

 red with increase in the thickness of the water column. 



As M. Billon-Daguerre wished to use light of very short 

 wave-length, he employed a vacuum tube filled with 

 hydrogen. This substance is known to give a strong 

 spectrum in the region between A. 1650 and \ 1030 ; it must 

 not be forgotten, however, that no lines can be ascribed to 

 it in the region between \ 2000 and \ 1650. Thus any 

 action due to the radiation from the vacuum tube filled with 

 hydrogen must be confined to a layer of water so thin that 

 light of wave-lengths shorter than A. 1650 can penetrate it. 

 Judging from my experiments, such a layer must be very 

 thin indeed. 



Several investigators have used the mercury arc in 

 quartz as a source of light in sterilisation experiments. 

 There are two facts which it may be interesting to mention 

 in this connection. In the first place, fused quartz two 

 .millimeires thick is somewhat transparent so far as \ 1500; 

 the transparency falls off rapidly with increasing thickness. 

 In the second place, no lines more refrangible than the 

 strong line at \ 1850 are known in the spectrum of mer- 

 cury. In this second statement my own observations are 

 confirmed by a recent investigation of Dr. Handke. 



Theodore Lyman. 



Jefferson Laboratory, Harvard University, July S. 



Elemental Weight Accurately a Function of the 

 Volution of Best Space-svmmetry Ratios. 



It ts a fact little known, but of the first magnitude, that 

 equ.il splirres or corpuscles cannot in space, as in one 

 plane, bf distributed at equal mutual distances. Tetra- 

 llcdra, the four points of which alone are all mutually 

 tquidistant, cannot be packed so as to fill space, as their 

 face-angles to fill one plane.' Icosahedral diffusion, with 

 a central sphere, nearly achieves this, but by a cramping of 

 the central point in the ratio i : i>osi460". 



Free magnetic needles in water, say five in number, may 

 fall into position either thus : 



Their energies are a fixed quantity ; so that, though they 

 will assume either position, they are stabler in position 

 (a), because here, on the whole, the lines are more equi- 

 distant ; but (b) might become equally stable if each needle 

 were a vortex possessing an energy v, capable, under heat 

 and cold, of adapting itself to changed environment by 



cumulative indraught and outdraught, i.e. v '". 



In one plane, equal spheres being equitriangularly 

 arranged, each sphere forms a centre capable of supporting, 

 by surface tension, an equal number of spheres around it. 

 In space, the nearest approach to this perfect equilibrium 

 is by means of the five best-symmetries, or so-called regular 

 solids, w-hereof three dominate elemental crystals.' .^Mke 



1 See liarlow anH Pope, Chemical Society Transactions, 1007, vol. xci., 

 - Retgers, Zeiisch. phvsical Chem., 1894, xiv., I. 



as to points, faces, edge-lines, and circum-radial lines, 

 these five contain only the factors 2 and 3 (crystalline) and 

 5 (non-crystalline), greatly complicated, however, by the 

 last of these ; 



Crystalline 



Non-crystalline 



: /lex \'2 : I ; let v'S : 3 ; oct V4 : 3 



■f)-' '%/<- 



3 ) 



Now^ the problem of the volutional interconversion (on 



the principle v -^'j of the three first ratios 2, 3, and 5, 

 yields to a simple and highly accurate solution," whereas 

 adding the two last, ic and do, the solution becomes com- 

 plex ; but, on the lines of the simple interconversion, there 

 are contained several approximate interconversions with 

 ic and do, the errors of which are the precise weights of 

 H' ■ ■ ' by different syntheses : 



or oit- 

 teP 

 hex' 



or $,'/:«- 

 or/cxH (i) 



- ihex" 

 or V^ 



or rVx H (ii) 



\'5 

 or V3 ;V X H' (iii) 

 or ^/2 ic- X H* itiy) 

 ,0(=t6xa!o' 



or 'J- 



< oct- 



(V) 



The numbers (i), (ii), (iii), (iv) refer to Morley's four 

 fxperimental weights of H,' which the formula hits pre- 

 :isely : 



(i) H.2: O mean =10076 1 

 (ii) H2:0 max. = 1-00777 



(iii) Gravimetry mean i '00762 

 (iv) II.,: H,,0 ,, 1-00765 



Two basal equations are here involved, 



(4/j)'=^2rax5r and 2'(ll3,)'' = f' -^VS"— 



7 and 12 being severally the combinable group and series 

 numbers of the table. The main equation {\hrees strong) 

 appears accurate to some 50 decimal points ; the secondary 

 Cfives strong) rather less so. They meet at s"'5='v'2; 

 with an error of 0-00016, the crux of the hydrogen ranges. 

 Their great accuracy points to a profound numeric and 

 geometric principle. Hex, hex-'oct-, (Src, compensation- 

 vortices cannot evolve to their 6th and gth roots without 

 developing hydrogen, and thereupon compensating ic, <5i-c. ; 

 and, inversely, ic, <S>-c., cannot involve to their 6th and "jth 

 povaers without ultimately throwing off hydrogen and com- 

 pensating hex, hex-ioct', <S~f. 



(i) The compensation-vortex at the end of the cubic (or 

 let Oct) edge-line, pulls, as required, by sj 2 : 1 against the 

 circum-cube radius. This crystalline symmetry being dis- 

 turbed by heat, the vortex unravels or evolves to its 6th 

 root, travelling down the line to the point marked i/v. It 

 there precisely compensates the icosahedral edge : circum- 



1 T,-t, ie.r, oct, ic, and do here stand for the ratio;, or the weights com- 

 oensating the ratios, edge-lhu •.circtim-radins {i.e. the radius of a circum- 

 scribed sphere! severally of the regular tetrahedon, cube, octahedron, 

 icosabedron and dodecahedron. 



2 A log-algebraic problem of eight terms unknown, it was soluble only by 

 reference to philosophical considerations anterior to those now discussed. 



3 Morley, confirmed by Thomsen, Keiser. Guye and Mallet. See Inter- 

 national Committee's Report. Chemical A'trus, February 12, 1897, May 5, 

 1899. June It, 1897, and May 12, 1903 ; or Freund's " Chemical Compoii- 



lion," 1904, p. 220. 



NO. 2125, VOL. 84] 



