Wov. 13, 1884] 



NA TURE 



3^ 



3Se s 



3 Tc, 



jCd 



3ALCI 



8AljBr 3 



3 Fe,Cl, 



3Ta 4 Cl, 



3Nb3Cl 5 



3 Nb 3 C I 



3Zr 4 Cl 4 



3 H.\ ; ,C1 

 3H 4 N 8 Br 

 3H1N3] 



.H5CI = 

 3H J X i ClHgCl = 

 3H 4 N 3 IH 

 3C1 

 3 Br 

 31 

 3Hg 

 3HgCl 



a ,s 

 •3P3CI3 



3A .' i . 



3 Bi 3 Cl 3 

 3PbCl 



o 4-433 



500 & 1040 4 -425 

 10-529 

 564 io-6 



S - S4i6 



1420 I 5-68 



9-0513 

 1 390 & 1439 9-04 

 3-9253 

 1040 3-94 



9-35I4 

 350 & 440 9-348 

 1S772 

 440 iS-62 



"'3S34 

 440 u-395 

 9-836 

 35° 



35° 



440 & 860 



440 



350 & 1040 

 S60 

 440 



35° 

 440 

 350 



9-6 



9 -520s 



9'6 



7-654 



7-SS 



8-0815 



S-i5 



8-3085 



8-2i 



8-35 Mitscherlich 



0-9294 



1-005 



1-7144 



171 



2-5457 



2-59 



i'4>43 



1-44 



3'3I34 

 35 



6-546 



6-49 



2 '47 



2-47 Berzelius 



5-611 



5-54 Mitscherlich 



S-935S 



8-Sq V. M 



7-0655 



7-03 Mitscherlich 



9-536 



9-8 Mitscherlich 

 I3-854 

 1 5 85 Mitscherlich 



7-758 



7-67 V. & C. Meyer 



4-814 



4-85 Mitscherlich 



4-875 Dumas 



6-3383 



6'3 1 ' 

 11-1871 

 11-16 Jacquelain 



9-536 



The agreement in all cases is such that, considering the diffi- 

 culties with which the determination of vapour-densities is 

 attended, it is not likely that other atomic weights could he 

 chosen to obtain like good results. If now the weights in 

 column / are taken to be the weights of a single atom for each 

 element in a certain solid or liquid state, the percentages of 

 oxygen in the following chlorates agree closely with the values 

 found by experiment, 1 to wit : — 



The agreement in these instances is as good as with the 

 adopted weights 1 but it is complete also in the following cases, 

 in which there are great discrepancies with the prevailing atomic 



weights : — 



iooPtCL.KCl contain 69-362 PtCU and 30-638 KC1 



j 69-417 ,, 30'5§3 ) ,, Berzelius 



J 69-318 ,, 30-682 \ ,, Seubert 



Mean 69-3675 

 yield 



30-6325 „ 

 117-825 AgCl 

 117-9606 ,, Seubert 



1 Proc. Roy. Soc. xxvii. p. 427. 



: Joicrn. Cheat. Soc. v. p. 69. 



The agreement of the mean of -the percentages of Berzelius 

 and Seubert with the calculated values is complete; the dis- 

 crepancy between the amounts of silver chloride is small and 

 within the limits of errors of observation. But the percentage- 

 of platinum and chlorine in PtCL arrived at by the two experi- 

 menters are widely different, viz. : — 



40-424 Pt ; 28-993 CI Berzelius 

 40-107 ,, ; 29-211 ,, Seubert 



The true weight of the chlorine follows from Seubert's analysis 

 of the ammonium salt — 



IooH 4 N 3 PtCl 3 yield I94'954 (AgCl) 3 

 Seubert obtained 192-846 ,, 



His rate between the silver chloride ) pt = I95 . 002 C i arke 



and the potassium salt gives ... \ 

 His rate between the silver chloride j =196-871 , ; 



and the ammonium salt gives ... ( " 



the latter rate is therefore at fault, and 100 parts of the ammonium 

 salt correspond to 194-694 AgCl, if the rate is the same as with 

 the potassium salt ; the difference between this number and 

 194-954 is within the limits of errors of observation. The rate 



- I0 ° XfAgCU, gives H 4 N 3 ClPtCL 2 = 70-84883, and the 

 I94-954 



rate 69 '^ X KC1 gives PtCl., = 53'95833 i H 4 N 3 C1 is there- 

 fore 16-8905, and as the weight of H 4 N 3 is not in doubt and 

 = 5-74468, CI is = 11-14583, as in column /. With this weighl 

 of chlorine all discrepancies disappear, while the weights recal- 

 culated from the same data vary between Pt = I94'3'4 a°<> 

 196-871. It is moreover minutely confirmed by the results 

 obtained from all the other elements of the same group. 



iooOsCl ? KCl ) ro226 Qs ; 28-5027 CI ; 3o'4747 KC1 

 contain \ ^ 



40-638 ,, 28-9024 ,, 30-4596 „ Berzehu 



Berzelius's percentage of chlorine is again too large, very nearly 

 to the same extent as the chlorine found by him in the potassio- 

 platinum chloride, while the percentage of the potassium chloride 

 is very exact. 



1 The experimental values are tl 

 (" Smithsonian Miscell. Coll.," vol. 



! recalculated by Prof. F. W. Clarke 



