277 On the Atomic Weight of Strontium .
By Sir Edward Thorpe , C.B. , LL. D. , F.R.S. , and Arthur Gordon Francis , .
B.Sc. , F.I.C. ( Received November 18 , 1909 , \#151 ; Read January 13 , 1910 .
) In the Bakerian Lecture for 1907 , " On the Atomic Weight of Radium , " * one of us described a rapid and effective method of separating the clear supernatant solution above a precipitate , say of silver chloride or silver bromide , which obviated the necessity of employing any of the ordinary methods of filtration and otherwise greatly simplified the manipulative process .
As the contrivance was found convenient in practice and seemed to admit of a high degree of accuracy , we have thought it worth while to make use of it in a redetermination of the atomic weight of strontium which should seek to conform to the standard of precision prescribed by modern procedure in atomic weight estimations .
Determinations of the atomic weight of strontium have been made by Stromeyer ( 1816 ) , f Rose ( 1816 ) , \Salv6tat ( 1843 ) , S Pelouze ( 1845 ) , || Marignac ( 1858),1T Dumas ( 1859 ) , ** and Richards ( 1894-5 and 1905).ft With the exception of those of Richards these estimations have no claim to great accuracy , and are therefore only of historical interest .
Richard 's final values as re-calculated by us when Ag = 107-880 , Cl = 35-460 and Br = 79-916 Ratio .
Series .
Mean .
2Ag : SrBr2 .
II 87 627 87*629 III 87-631J 2AgBr : SrBr2 ... .
I 87-621\87\lt ; 620 II 87*619 J 2Ag : SrCl2 The mean of the whole is Sr = 87'62 , which is the value adopted in the last report ( 1910 ) of the International Committee on Atomic Weights .
* ' Roy .
Soc. Proc. , ' A , vol. 80 , p. 298 .
+ ' Schweig .
J. , ' vol. 19 , p. 228 ; also Meyer and K. Seubert 's ' Atomgewichte , ' p. 123 .
1 ' Pogg .
Ann. , ' vol. 8 , p. 189 .
S ' Count .
Rend .
, ' vol. 17 , p. 318 .
|| 'Compt .
Rend .
, ' vol. 20 , p. 1047 ; and ' Liebig 's Ann. ' ( 1845 ) , vol. ' 56 , p. 204 .
TT ' Liebig 's Ann. , ' vol. 106 , p. 168 ; and ' N. Arch. Phys. Not .
, ' vol. 1 , p. 209 .
** ' Ann. Chim .
Phys. ' ( 1859 ) , vol. 55 , p. 191 ; and 'Liebig 's Ann. , ' vol. 113 , p. 34 .
++ ' Amer .
Acad. Proc. , ' vol. 30 , p. 369 ; and ibid. , vol. 40 , p. 603 .
VOL. LXXXIII.\#151 ; A. * rr 278 Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. 18 , We have sought to establish the atomic weight of strontium by means of the ratios:\#151 ; I. 2Ag : SrBr2 .
II .
2AgBr : SrBr2 .
III .
2Ag : SrCl2 .
IY .
2AgCl : SrCl2 .
Ratio IV has already been used by Rose ; Ratio III by P41ouze , Marignac , Dumas and Richards ; Ratios I and II by Richards alone .
Preparation and Purification of Materials .
All the materials needed in the course of the determinative operations were carefully purified .
The water was twice distilled from a metallic still , once from dilute alkaline permanganate and the middle third only was taken .
The hydrochloric acid and nitric acid were distilled from a platinum apparatus and the middle thirds only collected .
The red phosphorus , phosphoric oxide , sulphuric acid , etc. , were also treated by appropriate methods to ensure their freedom from any substance which would affect their employment .
All the platinum vessels were cleansed by being heated with acid potassium sulphate , washed with distilled water and treated with hot nitric and hydrochloric acid several times in succession .
The silver employed was prepared by reducing silver chloride ( obtained from the purest commercial silver nitrate ) by a boiling solution of potash and lactose , repeatedly washing the metallic silver with dilute aqueous ammonia and hot water , and fusing it when dry , in small quantities at a time , on a block of lime , the buttons being allowed to solidify in a reducing flame .
The buttons , which were of varying size , were scrubbed with a clean stiff brush and shaken together in a bottle with successive portions of dilute nitric acid to remove any trace of adhering lime and afterwards etched with nitric acid of specific gravity 1*2 .
After being again washed with pure water and dried , they were sorted into three sizes for convenience of weighing .
Some of the larger buttons were rolled between silver rollers into thin foil which was again washed and etched .
The silver was preserved in bottles kept in a desiccator containing fused potash .
As regards its purity , we have assumed that it was of the same character as that employed by one of us in connection with the determination of the atomic weights of gold* and which had been prepared in precisely the same manner .
This silver was estimated by the late Chief Assayer of the Mint , Sir William Roberts-Austen , after a series of very careful trials , to contain * 'Chem .
Soc. Journ. ' ( 1887 ) , vol. 51 , p. 569 .
On the Atomic Weight of Strontium .
1909 .
] 0 99994 gramme of real silver per gramme .
Exactly the same value is obtained from our present determinations if we compare the aggregate amount of silver chloride formed from the known weight of the silver used with the amounts calculated from Ag 107'88 and Cl 35*46 which are considered by the International Committee as the most probable values of these atomic weights .
Preparation of the Strontium Salts .
Strontium Chloride.\#151 ; About 500 grammes of commercially pure crystallised strontium chloride , obtained from Kahlbaum , was subjected to a systematic recrystallisation from water on the same principle as that adopted in the isolation of radium chloride.* The series consisted of five portions each and " heads " and " tails " were taken off from alternate series .
As the strontium salt is readily soluble , it is very necessary to free each crop of crystals from the mother liquor , as far as practicable , and this was effected in a small centrifugal apparatus .
After 30 series , involving 160 re-crystallisations , about 107 grammes of purified material was obtained .
This was still further purified by fractional precipitation from aqueous solution by means of pure hydrochloric acid , and the crystals so obtained were dried centrifugally as before .
The process was repeated eight times and four portions were eventually obtained .
These were separately dissolved in water , filtered under cover , and the solutions mixed with 1 c.c. of pure hydrochloric acid , and evaporated to dryness under a glass shade .
The mass was then dried at 200 ' in an oven heated by an alcohol flame .
Fractions II and III were employed for the determinations .
Strontium Bromide.\#151 ; 500 grammes of commercially pure strontium bromide was systematically recrystallised and " centrifuged " in precisely the same manner as the chloride .
In all there were 39 series involving 209 crystallisations .
After the 21st , 27th , 33rd , and 39th series the crystals of the middle fractions were " whirled " until dry , dissolved in water , the solutions filtered into a platinum vessel and evaporated to dryness under a glass shade as before .
The four portions were dried at 160 ' to 170 ' in an oven heated by an alcohol flame .
The fraction obtained from the 39th series was assumed to be the purest and it was preferred therefore for the determinations .
The purified and dried fractions of the two salts were preserved in stoppered bottles in a desiccator containing phosphoric oxide until wanted for use .
The specific gravities of the fused salts , obtained by displacement of * Bakerian Lecture , 1907 , loc. cit. U 2 Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. 18 , paraffin oil , were : SrCl2 3'051 ; SrBr2 4'201 .
Bichards ( .cit .
) obtained 4'216 for the bromide and adopted Schroder 's value of 3'05 for the chloride .
A special laboratory was set apart for the determinative work and this had attached to it a dark room in which the manipulation of the silver halides was done .
The balance was a new 10-cm .
short beam instrument , fitted with stirrup pans , made by Oertling , and was sensitive at its maximum load to l/ 50th milligramme .
To check tremor it was mounted on packets of filter-paper in the same manner as that used in the determinations of the atomic weight of radium .
The weights , of gilded brass , also by Oertling , were carefully standardised , the sensibility of the balance being determined for varying loads up to 100 grammes .
The object to be weighed was invariably tared by a precisely similar vessel of as far as possible the same weight and dimensions and both vessels were uniformly treated in exactly the same way .
The weighings were made by vibrations , observed through a lens , the zero point being ascertained before and after each weighing .
The corrections added to each gramme of the substance weighed were :\#151 ; Silver ( sp. gr. 10'5 ) ... ... ... . .
\#151 ; 0-000031 Silver chloride ( sp. gr. 5'59 ) ... + 0'000070 Silver bromide ( sp. gr. 6'47 ) ... .
-t-0'000041 Strontium chloride ( sp. gr. 3'05 ) ... 4-0000249 Strontium bromide ( sp. gr. 4'2 ) ... + 0000141 Guye and Zachariades , * have recently drawn attention to the fact already pointed out by Marignac in 1843 , that if the substances weighed are in the state of fine powder , the ordinary method of reducing weights to a vacuum gives a correction in excess of the truth .
As the materials actually weighed in our determinations were invariably fused , and were usually in compact masses , we think that the error due to this circumstance may be considered as practically negligible .
Determination of the Ratios SrBr2 : Ag2 and SrBr2 : 2AgBr .
An unweighed quantity of the purified and dried strontium bromide , about 2 grammes in amount , was placed in a platinum boat contained in a silica tube provided with ground-in connections , and , after displacing the air with dry nitrogen , gradually heated , first in the stream of nitrogen and then in a stream of dry hydrogen bromide , until the bromide fused .
The fused salt was allowed to solidify in a current of dry nitrogen which was then displaced by dry air and the bromide transferred whilst still warm , under conditions which precluded the absorption of atmospheric moisture , to a tared stoppered 50 c.c. * ' Compt .
Bend .
, ' vol. 149 , p. 593 ( 1909 ) .
On the Atomic Weight of Strontium .
1909 .
] flask and the exact weight of the anhydrous strontium bromide ascertained with all known precautions .
The strontium bromide was next dissolved in pure recently boiled water , the solution acidulated with nitric acid , and a determinate quantity of silver , usually about 1 to 2 milligrammes in excess , dissolved in nitric acid , was added to it under non-actinic light and the excess of silver then ascertained by titration by the Gay-Lussac method , using a solution of strontium bromide of known strength .
After the titration was finished , the clear supernatant liquid was removed by the method already referred to , and the precipitate of silver bromide washed , dried , fused , and weighed .
As regards details , the hydrogen bromide was made by dropping concentrated hydrobromic acid solution on a mixture of sand , red phosphorus , and phosphoric oxide , free from phosphorous oxide , and was further dried by passing through tubes containing a mixture of red phosphorus and pure phosphoric oxide .
The object of adding sand to the mixture was to moderate the violence of the production of the hydrogen bromide .
The fused strontium bromide was invariably colourless and its solution in water was perfectly clear and was neutral to litmus and phenol-phtlialein .
The weighed quantity of silver was brought into the burette seen in the figure together with a slight excess of a solution of pure nitric acid of known strength , the bulbs containing water were placed in position , and the solution of the metal started and maintained by a gentle heat .
The clear solution of the silver nitrate was then introduced drop by drop into the solution of the strontium bromide , kept in gentle rotation , through the tap , lubricated with pure water , and then the bulb-tube was turned on its axis so that its contents were emptied into the burette and the wash-water was transferred to the strontium solution .
The whole apparatus was then repeatedly rinsed with successive small quantities of water , and the washings delivered by means of the tap into the solution of the strontium salt .
After being well shaken the flask was allowed to stand in the dark room for about 18 hours before the titration was proceeded with .
For the final adjustment two solutions were used:\#151 ; ( 1 ) A solution of strontium bromide , made by dissolving 0*2640 gramme of the pure fused salt in 500 c.c. of water at 16'*5 ; 1 c.c. of this solution contained therefore 0*000528 gramme SrBr2 .
( 2 ) A solution of silver nitrate made by dissolving 0*2338 gramme of silver in a slight excess of nitric acid and diluting the liquid to 500 c.c. at 16'*5 ; 1 c.c. of this solution contained 0*0004676 gramme Ag .
Hence the solutions were very nearly equivalent .
The solutions as used were delivered from small ( 5 c.c. ) burettes carefully calibrated .
The excess of silver ( 1 to 2 milligrammes ) in the strontium solution was Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. 18J now ascertained in the usual manner , the operations being carried out in non-actinic light in the dark room .
As the end point was reached care was taken to allow ample time for the appearance of the last traces of the cloud .
When no further precipitation was noticed the liquid was back-titrated with the silver nitrate solution .
Finally when complete , an excess of about 3 c.c. of silver solution was added and the flask allowed to remain for about 18 hours in a dark room .
\#166 ; ... ... ... ... The clear liquid was then drawn over by the capillary syphon in the manner already described ( loc. cit. ) , and the precipitate of silver bromide washed by seven successive additions , first of very dilute nitric acid , and then of cold water .
The solution , containing mainly strontium nitrate , and the washing water were collected in a graduated cylinder , and their aggregate volume noted .
1909 .
] On the Atomic Weight of Strontium .
Assuming that 2 grammes of strontium bromide were used , equivalent to 1*71 gramme strontium nitrate in the solution over the precipitated silver bromide , that this solution measured 30 c.c. and that 27 e.e. were drawn off at each operation , 30 c.c. of washing water being uniformly added , then there would be left\#151 ; After the first drawing off " 1st washing ... ... ... . .
\#187 ; ... ... . .
J ) ... ... 5 ?
... . .
V ... . .
0T710 gramme Sr(N03)2 0-0171 0-00171 0-000171 0-0000171 0-0000017 0-0000002 2nd 3rd 4th 5th 6 th that is after the 4th washing the amount of any admixed strontium nitrate would be inappreciable by the balance .
The washed silver bromide was then dried , first at 110 ' , then at 160 ' , and was fused by placing the flask containing it on a platinum tray heated by an alcohol flame .
The tare was similarly treated , and both flasks were placed side by side in a desiccator containing phosphoric oxide for 15 to 20 hours before being weighed .
Refusion rarely made any difference in the weight of the silver bromide ; in no case did it exceed 0"02 of a milligramme .
The fused silver salt was always perfectly transparent , and of a faint yellow colour .
The weight needs a small additive correction on account of the solubility of silver bromide .
Determinations by electrical conductivity measurements of the solubility of silver bromide in water at ordinary temperatures have been made by Kohlrausch and Rose , * W. Bottger , f and Kohlrausch .
J The results are not very concordant , but the mean amount may be taken , without sensible error , as 0-2 milligramme per litre .
We assume that the amount is not materially affected by the presence of free nitric acid or of strontium nitrate .
In any case the correction is very much smaller in amount than for silver chloride , and has only a very slight influence on the value of the atomic weight .
The following example is given to show the character of the data obtained , and the mode of calculation by which the resultant atomic weight was deduced :\#151 ; * ' Zeit .
f. Phys. Chem. ' ( 1893 ) , vol. 12 , p. 240 .
t Ibid. ( 1903 ) , vol. 46 , p. 602 .
t Ibid. ( 1904 ) , vol. 50 , p. 356 .
Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. 18 , Grammes .
Difference in weight of flasks in vacuo 04 8687 Total weight in vacuo of flask and strontium bromide ... T89142 Nett weight in vacuo of fused strontium bromide ... ... .
1*70455 Weight of silver in vacuo ... ... ... ... ... ... ... ... ... . .
T48708 Titration required 1*9 c.c. SrBr2 solution .
Total added , 22 c.c. Burette correction = 0*003 milligramme SrBr2 .
Back titration required 0*35 c.c. AgNCh solution .
Burette correction , nil .
Then 1*48708 gramme Ag = 1*70455 + 1*9 x 0*000528 gramme SrBr2 = 1*70555 gramme SrBr2 , and 1*48708 gramme+ 0*35 x 0*0004676 = 1*48724 gramme Ag = 1*70455 + 2*2 x 0*000528 SrBr2 = 1*70571 gramme SrBr2 , of which the mean is 1*48716 gramme Ag = 1*70563 gramme SrBr2 .
Corrected on the assumption that 1 gramme of the metallic silver contains 0*99994 gramme real silver we have\#151 ; 1*48707 gramme Ag is equivalent to 1*70563 gramme SrBr2 , from which Sr = 87*639 .
After drawing over , total volume of wash water = 212 c.c. Correction for dissolved silver bromide = 0*04 milligramme .
Corrected weight of silver bromide in vacuo = 2*58862 .
Then 2*58866 grammes AgBr is equivalent to 1*70571 gramme SrBr2 , from which Sr = 87*651 .
The following are the results obtained from strontium bromide :\#151 ; Series A.\#151 ; Ratio SrBr2 : 2Ag .
No. of experiment .
Wt .
of SrBr2 .
Wt .
of Ag .
SrBrj 2Ag ' grammes .
grammes .
3 1 -77884 1 -55073 1 -1471 5 1 -86109 1 -62260 1-1470 6 1 -85254 1 -61511 1.1470 7 1 -73801 1 -51534 1 -1470 9 1 -85787 1 -61994 1 -1469 10 1 -70563 1 -48707 1 -1470 Mean 1 -1470 Mol. wt. of SrBr2 = 247*477 .
At .
wt. of Sr = 87*645 .
1909 .
] On the Atomic Weight of Strontium .
285 Series B\#151 ; Batio SrBr2 : 2AgBr .
No. of experiment .
Wt .
of SrBr2 .
Wt .
of AgBr .
SrBr2 2AgBr* 5 grammes .
1 -86112 grammes .
2 -82438 0 -65895 6 1 -85261 2 -81155 0 -65893 7 1 -73807 2 -63762 0 -65895 9 1 -85798 2 -81999 0 -65886 10 1 -70571 2 -58866 0 -65892 Mean 0 -65892 Mol. wt. of SrBr2 = 247'485 .
At .
wt. of Sr = 87*653 .
Determination of the Ratios SrCl2 : Ag2 and SrCl2 : 2 AgCl .
The general method of procedure as regards strontium chloride was similar to that described in the case of the bromide .
A quantity of the pure dry chloride was heated , first in a stream of nitrogen and then in a stream of hydrogen chloride , made by dropping pure sulphuric acid upon pure sodium chloride , until the strontium chloride fused , when the hydrogen chloride was displaced by nitrogen and subsequently by dry air .
The fused chloride was ice-like in appearance , and dissolved in water to a perfectly clear and neutral solution .
The rest of the procedure was precisely similar to that employed in the case of the bromide .
The solution of strontium chloride used in the titrations contained 0*0003384 gramme SrCl2 per cubic centimetre ; the silver solution was the same as that previously employed .
The fused silver chloride was always transparent and nearly colourless .
Refusion never affected the weight by more than 0*02 milligramme .
The weight of the silver chloride obtained needed an additive correction for the amount of the salt present in the solution and wash water drawn off .
Determinations of the solubility of silver chloride in water have been made by Kohlrausch and Rose* Bottger , *f and Kohlrausch| with fairly concordant results .
RichardsS has shown that the amounts are not sensibly affected by the presence of nitric acid , and according to Mulder|| the nitrates of the alkalis and alkaline earths are without influence at ordinary temperatures .
* ' Zeit .
Phys. Chem. ' ( 1893 ) , vol. 12 , p. 242 .
+ Ibid. ( 1903 ) , vol. 46 , p. 602 .
tIbid .
( 1904 ) , vol. 50 , p. 356 .
S \#171 ; Amer .
Chem. Soc. Journ. , ' 1905 , p. 486 .
|| Seidell 's ' Solubilities of Inorganic and Organic Substances , ' p. 285 ; also Coney 's ' Dictionary of Solubilities , ' pp. 372\#151 ; 3 .
Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. The mean amount at ordinary temperatures ( 18 to 20 ' ) deduced from the various observations is T52 milligrammes per litre , and this value has been made use of as the basis of the correction .
The results obtained with the strontium chloride are as follows:\#151 ; ; i Series C.-\#151 ; Eatio SrCl2 : 2Ag .
No. of experiment .
Wt .
of SrCl2 .
Wt .
of Ag .
SrCU w 10 grammes .
1 -64759 | grammes .
2 -24203 0 -73486 11 1 -66352 2 26356 073491 12 1 -53462 2 -08817 0 -73491 13 1 .
64619 2 -24011 0 -73487 14 1 -76u06 2 -39486 0 -73493 15 1 -56224 2 -12572 0 -73492 Mean 0-73490 #0 ' Mol. wt. of SrCl2 = 158-562 .
At .
wt. of Sr = 87'642 .
I if Series D.\#151 ; Eatio SrCl2 : 2AgCl .
No. of experiment .
Wt .
of SrCl2 . !
Wt .
of AgCl .
1 SrClo 2AgCl ' 10 grammes .
1 -64764 grammes .
2-97899 0 -55309 11 1 -66357 3 -00762 0 -55312 12 1 -53467 2 -77446 0-55314 13 1 -64624 2 -97653 0-55307 14 1 -76010 3 -18202 0 -55314 Mean 0-55311 Mol. wt. of SrCl2 = 168-565 .
At .
wt. of Sr = 87'645 .
mina ' I atom ; In order to obtain an independent check upon these determinations of the atomic weight of strontium , two further series of estimations were made depending on the ratio of SrS04 to SrCl2 and SrBr2 respectively .
The sulphuric acid needed to convert the halides into strontium sulphate was prepared by distilling " pure " sulphuric acid in a platinum apparatus , the middle third only of the distillate being employed .
It was preserved in a platinum bottle .
The strontium salts used were portions of the material of which the preparation has already been described .
They were fused in a stream of hydrogen bromide and chloride respectively , and otherwise treated as in the preceding series .
On the Atomic Weight 1909 .
] The weighed portions of the fused salts were transferred to a short wide burette-shaped vessel of about 75 c.c. capacity , fitted with a fine delivery tube provided with a stop-cock , and were then dissolved in pure warm water .
About 1\ times the equivalent amount of sulphuric acid diluted with water was next placed in a tared porcelain crucible , and both solutions were heated nearly to the boiling point .
The solution of strontium salt was allowed to flow very slowly and in a thin stream into that of the sulphuric acid which was kept constantly agitated .
In this manner the strontium sulphate was precipitated as a fine granular powder .
If strong solutions of strontium salts are precipitated with dilute sulphuric acid in the cold , the strontium sulphate separates out as a gelatinous mass which , on heating , gradually shrinks and becomes granular .
By operating with hot solutions , in the manner described , the danger of occluding undecomposed halide in the gelatinous precipitate is practically obviated .
The contents of the crucible were next brought to dryness , first by means of superheated steam under a bell-jar , and subsequently in a glass air-bath heated to 220 ' until no further evolution of acid vapours could be detected .
The dried mass wras then gradually heated to bright redness , together with its tare for about two hours , in a Meker furnace fitted with a silica muffle .
When cold , the strontium sulphate was moistened with 5 c.c. of dilute sulphuric acid , dried and again heated with the tare for an hour in the muffle , and after standing for about 18 hours in the desiccator over phosphoric oxide weighed with the usual precautions .
Both the crucible and its tare were again heated in the muffle and re-weighed , but no alteration in weight could be detected .
In reducing the weights to a vacuum the specific gravity of strontium sulphate was taken as 3*91 , being the mean value of four concordant determinations made on the substance after having been heated to redness .
The atomic weight of S was taken as 32-07 .
The results are as follows\#151 ; Series E.\#151 ; SrBr2 : SrS04 .
1 No. of !
experiment .
Wt .
of SrBr2 .
Wt .
of SrS04 .
SrJBr-2 SrS04* grammes .
grammes .
i 7 *14570 5 *30466 1*3471 2 7 *64281 5 *67326 1 *3472 3 9 *86072 7 *32047 1 *3470 Mean 1*3471 At .
wt. of Sr = 87-629 .
Sir E. Thorpe and Mr. A. G. Francis .
[ Nov. 18 , Series E.\#151 ; SrCl2 : SrS04 .
No. of experiment .
Wt .
of SrCl2 .
Wt .
of SrS04 .
SPI \#171 ; * 1 op | 1 grammes .
7 *30246 grammes .
8 -46071 0-86310 2 8 -71628 10 -09868 0 -86311 3 8 *46493 9 -80743 0 -86311 4 8 -79502 10 -18957 0 -86314 Mean 0 -863115 At .
wt. of Sr == 87'661 .
This method hardly admits of the same degree of accuracy as the methods of Series A , B , C , and D. The probable error of the Series A , B , C , and I ) is 0*0016 , whereas in the Series E , F , it is 0*0107 , being much greater in E than in F. It is to be observed that even when using quantities up to 10 grammes in series E and F an error of 0T milligramme affects the value of the atomic weight to the extent of 0*01 unit .
Moreover , the vessels to be weighed are much larger in volume , and this affects the degree of accuracy with which weighings can be made .
The manipulative process , too , is more complicated , and is attended with more numerous sources of error .
As regards the effect of the occlusion of possibly undecomposed halide , it may be pointed out how this acts in the case of the two salts .
In the case of the bromide , since 2Br = S04 and 159*83 = 96*07 , the effect of undecomposed bromide would be to increase the apparent weight of the sulphate , and thereby increase the value of the atomic weight ; whereas , in the case of the chloride , since 2C1 = S04 and 70*92 = 96'07 , the effect would be to decrease the apparent weight of the sulphate , but this decrease would also increase the atomic weight .
It is conceivable , too , that some reduction of the strontium sulphate might have taken place owing to the difficulty of altogether excluding atmospheric organic dust , in spite of the care taken in the evaporation .
This dust would be charred by the sulphuric acid at 220 ' , and , on ignition , might reduce traces of the sulphate .
The second treatment with sulphuric acid was intended to remove any possible trace of sulphide in the weighed sulphate .
It should be stated also that the atmosphere of the Meker furnace was wholly oxidising , and no alteration of weight was found to occur as the result of successive heatings each of an hour 's duration .
The variations in the individual experiments are most probably mainly M- jjssa'M 1S#0 ' I ColW ' % On the Atomic Weight of Strontium .
1909 .
] due to mechanical loss , arising , in spite of every precaution , from " spitting " in the course of the evaporation and ignition .
As regards possible error in the calculation of the true weight of the finely divided strontium sulphate in , due to possible occlusion of air , we have no valid data for its estimation .
But considering the nature of this salt , the error due to this cause must be very small , if not altogether negligible , since the strontium sulphate after ignition was always in the form of a dense coherent cake .
Collecting the results , we have the following values for the atomic weight of strontium:\#151 ; From Series A . .
87-645 +0-0037 99 B . .
87-653 +0-0045 99 c . .
87-642 +0-0017 99 D . .
87-645 +0-0020 99 E . .
87-629 +0-021 99 F . .
87-661 +0-0078 99 A , B , C , D . .
87-646 +0-0016 99 E , F . .
87-645 +0-0107 99 A , B , C , D , E , F ... .
... 87-646 +0-0029 We adopt , then , 87"65 as the definite value for the atomic weight of strontium\#151 ; a number only 0'03 in excess of that already obtained by Richards , and given in the last Report of the International Committee .

