82 The Relative Atomic Weights of Nitrogen and Sulphur .
By F. P. Burt and F. L. Usher .
( Communicated by Sir William Bamsay , F.R.S. Received May 26 , \#151 ; Read June 30 , 1910 .
) ( From the Chemical Laboratory , University College , London .
) In 1905 , nitrogen sulphide was being prepared and used in large quantities at University College , Bristol , and it was due to a suggestion of Prof. Francis that it might prove a suitable substance for determining the relative atomic weights of its constituent elements , that the present research was undertaken .
Ho direct measurement of the ratio N : S has previously been made , and since the atomic weight of sulphur is still uncertain , a useful step is gained in acquiring an accurate knowledge of that ratio .
The compound U4S4 has the advantage of containing no other element .
Preparation and Purification of Nitrogen Sulphide .
All the nitrogen sulphide employed in this research was made in Bristol by Prof. Francis and Mr. 0 .
C. M. Davis , by passing dry ammonia into a benzene solution of sulphur chloride .
The main products of the reaction are nitrogen sulphide , nitrogen , and hydrogen chloride .
The sulphide which was deposited on concentrating the benzene solution contained traces of sulphur .
We used as our raw material some sulphide which had stood for many months in the air , and which contained a large proportion of free sulphur as well as ammonium salts .
The mixture was extracted with pure dry benzene in a Soxhlet apparatus , and the resulting solution contained all the nitrogen sulphide and relatively little sulphur ; on cooling , large needles of nitrogen sulphide were deposited , together with a little finely divided sulphur .
The crystals , after being twice washed with cold carbon disulphide , and then with benzene , were powdered on a porous the and placed in a desiccator containing calcium chloride and cocoanut charcoal .
Prof. Travers suggested that an attempt should be made to analyse the sulphide in such a way as to obtain free nitrogen , which could be measured in a constant volume gas thermometer , and with this end in view it was sought to decompose the substance by heating it vacuo .
It was found that at about 80 ' the sulphide began to volatilise without decomposition , being deposited again as a crystalline sublimate on the cool parts of the vessel .
If the sulphide was sublimed in vacuo over a roll of silver gauze , tightly packed in a vertical glass tube , decomposition was complete when The Relative Atomic Weights of Nitrogen and Sulphur .
83 the temperature of the silver was above 200 ' , though at temperatures between 100 ' and 120 ' the vapour was not decomposed to any appreciable extent .
This fact at once suggested a satisfactory method of freeing the sulphide from any traces of sulphur .
Blank experiments showed that if free sulphur was similarly sublimed in vacuo below silver gauze , the silver blackened evenly from below upwards , and no trace of sulphur could be detected on the clean glass above the gauze .
The powdered nitrogen sulphide , obtained as described above , was placed in quantities of about 1 gramme in a series of vertical glass tubes which could be exhausted by means of a mercury pump .
Above the sulphide in each tube a roll of silver gauze about 5 cm .
long was fitted .
A steam jacket surrounded each tube up to the level of the top of the silver gauze.* The nitrogen sulphide was then sublimed in vacuo , when it was deposited in crystals on the upper cool portions of the tube .
From 12 to 24 hours ' heating was necessary to sublime 1 grm. of the sulphide .
It was noticed that a deep blue transparent film gradually made its appearance on the walls of the sublimation tube some way above the yellow sulphide , and finally crept down to meet the latter if the heating was continued long enough .
The probable explanation of this curious phenomenon is discussed by one of us elsewhere.f The blue substance has been shown by analysis to be a sulphide of nitrogen , having the same percentage composition as the ordinary sulphide , but under these conditions it is formed in practically unweighable quantities .
During the process of sublimation the silver gauze was superficially coated with sulphide , and a small quantity of nitrogen was pumped off , amounting to a fraction of a cubic centimetre after 12 hours ' heating .
The traces of gas evolved had an odour of benzene , carbon disulphide , and sulphur dioxide ; the last was possibly due to traces of oxygen in the silver , while the others must have been derived from the solvents used .
We never succeeded in obtaining nitrogen sulphide that did not yield traces of benzene ( or other solvent ) on sublimation in vacuo unless that nitrogen sulphide had already been sublimed in vacuo .
At the end of the experiment the sublimation tubes contained a small white residue , consisting of fragments of filter paper and porous the .
The product obtained by sublimation over silver in vacuo consisted of large , orange-coloured crystals , which must necessarily have been free from liquid or gaseous inclusions .
When struck , or rapidly heated in air , they detonated violently .
Dr. Herbert Smith has kindly examined their crystalline form , * This was previously heated red-hot in the outer cone of a blow-pipe flame in order to remove oxidisable impurities .
t 'Trans .
Chem. Soc. , ' 1910 , vol. 97 , p. 1171 .
G 2 Messrs. F. P. Burt and F. L. Usher .
[ May 26 , and finds that they belong to the monoelinic system .
They are therefore of the same type as the crystals deposited from solution , examined by Artini.* The specific gravity of the sublimed sulphide , which was required in subsequent calculations , w*as determined by weighing in air and water at 20 ' in a small specific gravity bottle .
We obtained as the result of two determinations the values 2*248 and 2*235 , giving a mean of 2*24 , sufficiently accurate for a vacuum correction .
The specific gravity of the unsublimed sulphide was also determined ; we obtained the values 2*21 and 2*19 , giving a mean of 2*20 , which is in agreement with the results of other observers .
Nitrogen sulphide is so slowly hydrolysed by water that it is perfectly safe to use this liquid .
A large number of analyses were carried out by subliming nitrogen sulphide in vacuo over silver gauze heated to about 360 ' .
The silver gauze ( which was supplied by Messrs. Johnson and Matthey ) was freed from oxygen by heating to redness in a stream of hydrogen , and the occluded hydrogen was got rid of by prolonged exhaustion at the same temperature .
The gauze was finally exhausted from nitrogen at 360 ' .
As the results obtained were unsatisfactory , the apparatus is not described .
There was every reason to believe that the nitrogen sulphide was completely decomposed by the silver ; the line of demarcation between the silver sulphide and the unattacked silver was perfectly sharp and definite at the end of a determination ; moreover , the cool part of the reaction tube above the silver gauze would have shown the presence of unweighable traces of nitrogen sulphide or sulphur , and the glass was found to be always perfectly clean .
We are inclined to attribute the discrepant results to the occlusion of nitrogen by silver sulphide at the moment of its formation .
A similar phenomenon has been observed by Dr. R. W. Gray in the reduction of nitric oxide by metallic nickel.f If this is the correct explanation , it would have been necessary to fuse the silver sulphide , which would have involved manipulative difficulties of so serious a nature that we decided to devise some other method of analysis which should dispense with the employment of a metal .
Our first attempt was to replace the silver by tightly packed quartz wool , but we found that at 360 ' the quartz only partially decomposed the nitrogen sulphide .
We finally succeeded by using a quartz reaction tube , in which quartz wool was heated to bright redness by a naked flame .
* * Zeitschr .
fur Kryst .
, ' 1906 , vol. 42 , p. 68 .
t " tiber das Atomgewicht des Stickstoffs , " Inaug .
Dissert .
, Bonn , 1907 .
1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
85 Description of Apparatus and Details of Procedure .
The figure represents the apparatus in which our final analyses were carried out .
A weighed quantity of nitrogen sulphide in the small quartz bucket A was placed in the quartz reaction tube B ; a plug of quartz wool C , which had been treated in a manner to be subsequently described , was then pushed into the centre of the reaction tube , and the latter was attached to the greased push-join D. The Topler pump E could be put in communication c SCALE x 0 0925 either with the reaction tube or with the manometer head F by means of the three-way tap G. The capillary fall tube of the pump was turned up and sealed as an inserted join into the gas collector H. A capillary tube sealed to the upper end of H carried another three-way tap J leading on the one hand to the measuring cylinder K and on the other to the potash tube L , which could also be opened to the air through the third three-way tap M. The measuring cylinder K was connected by means of capillary tubing with the dead space N and manometer F. Before a determination was started Messrs. F. P. Burt and F. L. Usher .
[ May 26 , the dead-space , measuring cylinder , and connecting capillaries , after exhaustion by the pump , were filled with mercury from the reservoir 0 .
The reaction tube was then exhausted , the air being expelled through the collector and potash tube ; after exhaustion was complete , any traces of gas sticking at the end of the capillary fall tube or on the walls of H were collected by lowering the reservoir P with the tap J closed , until the mercury in the collector fell to the level of the inserted capillary ; the reservoir was then raised again , and any gas collected was expelled into the potash tube L and thence into the air .
The apparatus was now ready for a determination to be started .
A D-shaped roof of asbestos B was placed over the reaction tube so as to cover the quartz wool , and the last 8 cm .
of the reaction tube passed through a closely fitting circular aperture into a cubical asbestos oven S , which could be heated by an ordinary Bunsen burner .
The oven carried a thermometer , of which the bulb was in close proximity to the nitrogen sulphide in the tube .
A stream of cold water was arranged to drip on the junction D throughout the determination .
A large , roaring Bunsen flame was placed under the quartz wool in the reaction tube , and after an interval of about ten minutes the flame was lit under the asbestos oven .
The nitrogen sulphide was volatilised at a temperature which rose throughout the determination from 100 ' to 170 ' , and the pump was worked so as to keep the average pressure of nitrogen at a few millimetres of mercury .
The sulphur was deposited as a lemon-yellow crust in the neighbourhood of the push-join .
As soon as the collector H was full of gas , the latter was at once transferred to the measuring cylinder , and , at the end of the determination , the last traces of gas were collected in the way already described , and added to the main quantity in the measuring cylinder .
Mercury was finally taken through the tap J and set to a mark T on the capillary .
Measurement of the Nitrogen .
The volume of the measuring cylinder between two marks T , U , on the capillary stem was determined on two occasions with an interval of four years .
The earlier calibration was made with water , and the later with mercury ; the following are the values obtained\#151 ; Calibration with water ^ ^flo ' CTMean = 105,372 c.c. L(n ) 105-373 " J , , f ( i ) 105-367 c.c.T in Calibration with mercury i05-'70 J*Mean = 10o-369 c.c. The agreement obtained was very satisfactory , and since the small difference after a period of four years was within the limits of experimental 1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
87 error , the more recent value was adopted .
The capillary volume between the marks U and Y was calibrated with mercury , and amounted to 0T52 c.c. The lead-glass dead-space , manometer tube , millimetre scale , and the water-bath in which they were set up had already been used by Dr. Gray and one of us in a research on the compressibility of hydrogen chloride.* The volume of the dead-space was 1-275 c.c. To carry out a measurement of nitrogen , the measuring cylinder K was surrounded by a bath containing powdered ice and water , and the mercury was set exactly to a glass point in the dead-space .
The distance between the upper mercury meniscus in the manometer tube and the dead-space point was then determined by means of the scale suspended in the bath , fractions of a millimetre being measured by a telescope provided with a Hilger micrometer eyepiece .
For a fuller description of this part of the apparatus , the paper already referred to may be consulted .
The temperature of the mercury column was kept constant by a stream of water circulating through the bath , and was registered by two thermometers , one at the level of the dead-space and the other at the level of the upper meniscus .
In this way the pressure exerted by a known volume of nitrogen at a known temperature was obtained , and in order to - calculate its mass a knowledge of the density of the gas was all that was required .
We have taken the value 1'2514 grm. as the weight of a litre of nitrogen at 0 ' and 760 mm. in London .
The question of the density of nitrogen is considered at the end of the paper .
The nitrogen sulphide was weighed in a small quartz bucket , provided with a handle so that it could be lifted with a glass hook without contact with the fingers .
The weight of the bucket was about half a gramme .
It was carefully cleaned by treatment with hot chromic acid and subsequently heating to redness on a platinum wire , and was first weighed empty .
A suitable quantity of nitrogen sulphide was then introduced by means of a glass spatula and the bucket was weighed again , and then immediately lowered into the reaction tube .
While the weighings were in progress the reaction tube was cleaned with chromic acid , washed , dried , and finally heated red-hot .
Preparation of the Quartz Wool .
The quartz wool employed was obtained from the Silica Syndicate in the form of a hank of fibres with a mean diameter of about 0 01 mm. It was first of all necessary to " felt " this by hand so as to form a compact plug , * 'Trans .
Chem. Soc. , ' 1909 , vol. 95 , p. 1659 .
88 Messrs. F. P. Burt and F. L. Usher .
[ May 26 , and i , t was found that 3 grm. when so treated occupied a length of about 6 cm .
when tightly pressed into the reaction tube .
Such a plug was .calculated to contain 10f miles of fibre with a surface of over 5,000 sq .
cm .
It was realised that so large a surface might involve an appreciable correction for occluded air , and a series of measurements was made to elucidate the question .
After the process of " felting , " the wool was packed into an open-ended quartz tube\#151 ; of the same internal diameter as the reaction tube\#151 ; and heated for several hours to bright redness in a stream of dry air , freed from dust and carbon dioxide , and finally allowed to cool in the air-stream .
In this way any carbonaceous matter was oxidised and the wool was never again touched with the fingers .
From the open quartz tube it was pushed into the reaction tube by means of a glass rod .
The reaction tube was attached to the pump , exhausted in the cold , and then heated as in an actual analysis .
The small quantity of gas evolved was collected and measured in the calibrated capillary at the top of the collector H. When no more gas could be pumped off the tube was allowed to cool and air was admitted through a soda lime tube at the capillary W. The reaction tube was removed and most of the grease in the mouth was wiped off with a clean rag .
The tube was then held horizontally and the last traces of grease were burnt out of the neck by heating red-hot with a naked flame in a current of dust-free air .
In order to prevent backward diffusion of the products of combustion on to the quartz wool , a piece of quill tubing of silica , which conducted the air current , was pushed well inside the mouth of the tube beyond the heated zone.* As soon as the tube was cold the wool was transferred by the aid of a glass rod bent into the form of a corkscrew to the open-ended quartz tube and the latter was at once put into a desiccator containing potash .
In this way the wool underwent precisely the same manipulation and short exposure to the air of the laboratory that were inevitable between two actual analyses .
Two more blank experiments were made with the same wool in exactly the same manner as just described .
The volumes of air obtained in the three cases were:\#151 ; ( i ) 0-0173 c.c. ( ii ) 0-0099 c.c. ( iii ) 0-0056 c.c. The diminution in the volume of occluded air was due partly to a small mechanical loss of wool , occasioned in transference , and partly , perhaps , to a decrease in its occluding power , owing to the prolonged heating .
* After an analysis of nitrogen sulphide the mouth of the reaction tube contained not .only grease but sulphur , and this was burnt out with the grease in the manner described .
In order to keep the conditions as similar as possible , a piece of sulphur was placed in the neck of the reaction tube and burnt out with the grease in " occlusion " experiments , .on occasions when a sublimate was not already present .
1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
89 If the volumes of air obtained are plotted against the ordinal numbers of the experiment , the three points lie very nearly on a straight line .
In any series of analyses , therefore , it was only necessary to make two occlusion measurements , one at either end of the series , and to use a straight line interpolation for fixing the correction to be applied to the intermediate determinations .
It was found inconvenient in practice to start any series of experiments with a plug of quart/ wool occupying a greater length than 6 to 7 cm .
of the reaction tube , and this quantity only sufficed for about six experiments ( whether occlusion measurements or analyses ) partly because of the small constant loss , and partly because the continued heating and manipulation rendered the wool inelastic and powdery , when it would no longer form a tight plug in the reaction tube .
It will probably be conceded that no more air , that is , oxygen and nitrogen , would be washed out of the hot wool when nitrogen sulphide was being sublimed over it than would be pumped off by exhaustion in vacuo at the same temperature , though this would not be true for any denser gases which might find access to the wool .
Of such gases , the only two which could possibly have reached a significant concentration in the room in which we worked were carbon dioxide and sulphur dioxide , * and since it was impossible to avoid exposing the wool to the laboratory air for a short period between each experiment , it was essential to investigate the possibility of quantitative error from this source .
We may state at once that , so long as the wool was exposed to the laboratory air for indeterminate periods , we failed to obtain constant results for the N : S ratios , even when the necessary correction was made for occluded air , but it was subsequently discovered that if the nitrogen volumes in a series of determinations were treated with moist potash and then dried , a contraction occurred , which at once brought the ratio to a constant value .
The results of a large number of our earlier determinations might have been corrected , had we at that time suspected the presence of traces of gas absorbable by potash .
A point still to be considered was whether the air pumped off the hot wool in vacuo contained any gas absorbable by potash .
Since two separate corrections were made on the nitrogen , namely , that for air , which was fixed by the occlusion experiments , and that for carbon dioxide or sulphur dioxide , which was determined by measuring the contraction of the nitrogen on treatment with potash , the total correction would have been too large if any carbon dioxide had been measured as occluded air .
By exposing the * Neither ammonia nor water vapour need be considered , as both would be removed by the phosphorus pentoxide between the reaction tube and the pump .
Messrs. F. P. Burt and F. L. Usher .
[ May 20 , minute quantities of gas obtained in the occlusion experiments to potash , and then remeasuring them , a very small contraction of the order of 1 cu .
mm. was observed .
This could be neglected without appreciable error .
At an early stage in our experiments with a quartz reaction tube , it was discovered that a small quantity of hydrogen from the flame passed through the hot quartz ; moreover , when a similar flame was used , the quantities of hydrogen that entered the tube in unit time were remarkably constant .
This conclusion was arrived at in two different ways , and was repeatedly confirmed .
In the first place , it was noticed that it was impossible to pump the heated tube " tight , " gas continuing to collect in the pump at the rate of about 3*5 cu .
mm. per hour .
In order to apply a direct test for hydrogen , the gas obtained in a particular occlusion experiment was mixed with a little additional air , and transferred to a small gas burette provided with , platinum electrodes .
On passing a spark a contraction occurred :\#151 ; Volume before sparking = 0*065 c.c. Volume after sparking = 0*046 " Contraction = 0*019 " Therefore Volume of hydrogen = 0*019 x S = ( K)13 c.c. in 3f hours = 0*0036 c.c. per hour .
Values obtained by direct measurement of the eorfstant quantity of gas , , collected when the empty tube was heated , varied from ( M)033 to 0*0040 c.c. per hour .
For correcting our nitrogen volumes , we assumed the hydrogen to enter at the rate of 0*0036 c.c. per hour .
Treatment of Nitrogen with Potash .
After the initial measurement of nitrogen , the gas was taken , vid the collector , into the tube L. This tube had been prepared before it was set up , by melting a stick of caustic potash in it , and rotating it so as to cover the walls with a film of the fused alkali .
Before letting in the nitrogen , the mercury in L was taken down to the bottom of the tube , with the tap M closed , by lowering the reservoir Q ; any bubbles of gas caught in the potash film escaped into the vacuum so produced , and were collected and expelled into the air , when the mercury was taken up again .
This operation was always carried out twice .
A thread of water ( 30 or 40 cu .
mm. ) was now introduced through the capillary tube X , so as to moisten the potash , and the nitrogen was then transferred to U After standing over-night , the nitrogen was returned to the measuring cylinder , again with the same precaution of lowering the mercury , so as to obtain the last traces , and the 1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
91 mercury in the dead-space was taken below the level of the side-tube Y. By an alternate manipulation of the taps Z and Z ' , the gas was cautiously taken into the pump until the pressure in the measuring cylinder was reduced to a few centimetres of mercury .
Keeping the taps Z and Z ' open , a stroke of the pump was taken , so as to fill the collector H ; the tap J was then opened , so that gas flowed from the collector to the measuring cylinder , and thence over the phosphorus pentoxide hack into the pump chamber .
About 30 strokes were taken , so as to dry thoroughly not only the gas , but also the glass walls of the collector , measuring cylinder , and dead-space .
When this was accomplished mercury was allowed to rise in the dead-space , and seal the tap Z , which was then turned off .
The rest of the nitrogen was then returned to the measuring cylinder , by working the pump till exhaustion was complete ; the last traces were collected by lowering the reservoir P , and the mercury was set again to the mark T on the capillary .
A second measurement of the nitrogen was then made .
We found that the operation of circulating the gas through the system could be carried out without gain or loss ; on repeating the whole process with a quantity of nitrogen which had already been treated with potash and dried , we recovered the original volume to within 1 part in 50,000 .
In the two series of experiments of which the results are given in the following table , the quartz wool was treated in a precisely similar way .
The addition of the potash tube to the apparatus was made between the two series , so that the actual contraction of the nitrogen on treatment with potash was only measured in the second series .
Since , however , the ratios from Experiments 2b and 3b were in good agreement respectively before and after the gas had been treated with potash , and since , further , the uncorrected results of Experiments 2a , 3a , and 4a agreed with each other , and with the uncorrected results of 2b and 3b , we applied the correction determined in these two cases to the corresponding experiments in series A. Again , since the corrected result of Experiment 1b agreed with the corrected results of 2b and 3b , the same contraction correction actually determined in the case of 1b was applied to 1a .
In every series of analyses with a given specimen of quartz wool , the value from the first experiment was always higher than the values from the succeeding ones .
This was apparently due to the presence of a larger proportion of gas absorbable by potash , as indicated by the contraction measured in Experiment 1b .
When the history of the wool and the time of its exposure to laboratory air had also been the same , as in the two series given , the values from the first experiments also agreed very closely .
The values obtained from the second , third , and sometimes the fourth to Table of Eesults .
No. of experiment .
Nitrogen sulphide .
Yacuum correction .
Pressure of gas .
Yolume at N.T.P. Correction for hydrogen .
Occluded air .
Gas removed potash .
Corrected volume .
Density at observed pressure .
Weight of nitrogen .
Weight of sulphur by difference .
Eatio , N/ S. lA grin- 0 -469455 mgrm .
0-258 mm. 812 -73 c.c. 114 -162 c.c. 0-013 c.c. 0-019 c.c. 0-084 c.c. 114 -046 1 -25148 grm. 0 -142726 grm. 0 -326729 0 -43685 2a 0 -442787 0-243 766 -18 107 -624 0-012 0-017 0-018 107 -578 1 -25144 0 -134627 0 -308160 0 -43688 3a 0 -456326 0-249 789 -50 110 -902 o-oii 0-014 0-018 110 -859 1 -25146 0 -138736 0 -317590 0 -43684 4a 0 -470072 0-258 813 -29 114 -240 o-oio 0-012 0-018 114 -200 1 -25148 0 -142919 0 -327153 0-43686 1b 0 -466918 0-252 808 -41 113 -550 o-oio 0-014 0-084 113 -442 1 -25147 0 -141969 0 -324949 0--43690 2b 0 -491307 0-261 850 -07 119 -404 0-012 0-013 0-018 119 -361 1 -25150 0 -149380 0 -341927 0-43688 3b 0 -484307 0-265 837 -97 117 -704 0-011 o-oii 0-018 117-664 1 -25150 0 -147257 0 -337050 0 -43690 Mean N/ S 0 -43687 ' PvnhnKlp error \#177 ; 0-000006 1 * S Messrs. F. P. Burt and F. L. Usher .
[ May 26 , 1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
93 experiment in a series , were nearly constant .
Following these constant results came a much lower value , suggesting that some of the nitrogen sulphide had escaped decomposition , because of the diminution in quantity and efficiency of the quartz wool .
As a matter of fact , a low value could always be foretold , before the nitrogen was measured , by the appearance of the sulphur sublimate .
As long as decomposition was complete , the sulphur solidified at once on cooling , and the colour of the sublimate was a pale lemon-yellow ; if , on the other hand , complete resolution of the nitrogen sulphide into its elements had not been .
effected , the sulphur did not solidify quickly on cooling , and eventually exhibited darker patches of a red or brown colour .
A glance at the table on p. 92 will show that the three separate corrections made on the volumes are very small compared with the total nitrogen volumes ; moreover , the probable error on the correction itself is in each case very small .
In the case of the hydrogen , the experiments already cited indicate the order of this error to be less than a cubic millimetre for three hours heating , which was the average time occupied by a determination .
The largest correction for occluded air is less than 1/ 6000 of the total nitrogen volume , and the probable error on the correction itself is not greater than 5 per cent. Finally , the validity of applying a similar contraction correction in cases where it was not actually measured , is supported by many preliminary experiments which showed that quartz wool will take up from the air condensible gases in amount depending almost entirely on the time of exposure , and also by the very good agreement of the ratios among each other before this correction was made .
The agreement already referred to between the constant values on the one hand and the high initial values on the other is evident from the following list of ratios uncorrected for gases absorbed by potash:\#151 ; Experiment ... .
1a .
1b .
2a .
3a .
4a .
2b .
3b .
Ratio N/ S ... . .
0-43735 0-43740 0-43698 0-43695 0-43696 0-43699 0-43701 The following is a full calculation from the data in Experiment 1b :\#151 ; Weight of bucket + nitrogen sulphide Weight of bucket ... ... ... ... ... Weight of nitrogen sulphide ... ... ... ... ... ... ... ... ... ... ... ... ... .
0"466666 " Vacuum correction ( 17 ' and 758 mm. ) ... ... ... ... ... ... ... ... ... ... ... +0*000252 " 1-032683 grm. 0-566017 " Corrected weight of nitrogen sulphide 0-466918 Messrs. F. P. Burt and F. L. Usher .
[ May 26 , Nitrogen reading:\#151 ; Volume of measuring cylinder at 0 ' ... ... ... ... ... ... . .
Volume of dead-space , 1'275 c.c. Temperature of gas in dead-space , 8''83 Volume of gas in dead-space at 0 ' = \#151 ; ... ... . .
Volume of intermediate capillary , 0T52 c.c. Volume of gas in intermediate capillary at 0 ' = ^ ^28 105'369 c.c. 1-235 " 0-142 " Total volume at 0 ' and at observed pressure ... ... .
Pressure measurement:\#151 ; Micrometer readings .
Millimetres .
Distance from meniscus set 84 , 84 , 84 , 84 , 85 , 0'654 to point in dead-space to 85 : mean 84-33 scale-line 1 above it Distance of upper meniscus 636 , 634 , 636 , 635 : 4'924 below scale-line 815 mean 635'25 Uncorrected manometric height , 810-076 \#151 ; 0-346 = ... ... ... ... ... ... Correction for scale errors ... ... ... ... ... ... ... ... ... ... ... ... ... .
Contraction of scale from 16 ' to 8''9 ( = 7'1 x 0-000009 x 810 ) ... .
Correction of observed manometric height from 8 ' " 88 to 0 ' ... ... . .
Corrected pressure ... Volume of nitrogen at N.T.P. = U6'751 x 8Q8'405 = 8 760 Correction for hydrogen ... ... ... ... ... ... ... ... ... . .
Correction for occluded air ... ... ... ... ... ... ... ... .
Contraction on treatment with potash ... ... ... ... ... .
106-751 " Scale readings 0"346 mm. 810-076 " 809-730 " + 0-034 " - 0-052 " - 1-307 " 808-405 " 113"550 c.c. - 0-010 " - 0-014 " - 0-084 " Corrected volume of nitrogen at N.T.P ... ... ... ... 113-442 " Density at observed pressure = 1-25144 x 1-000027 = 1*25147 . . .
Weight of nitrogen = 113-442x0-00125147 = 0T41969 grm. Weight of nitrogen sulphide taken ... ... = 0-466918 " Weight of sulphur by difference ... ... ..= 0-324949 " Ratio N/ S = 0-436897 " Discussion of Errors.\#151 ; The question of the purity of the nitrogen sulphide has already been considered .
With regard to its stability in dry air we may remark that no change in weight in a specimen was detectable after several days ' standing in a desiccator ; moreover , it was found to be immaterial as regards the resulting ratio , whether we used nitrogen sulphide just removed from a sublimation tube , or crystals that had been kept for several days in dry air .
In order to ascertain whether there was any danger of losing nitrogen sulphide by volatilisation during the preliminary exhaustion of the reaction tube at the room temperature , a weighed quantity of the finely divided substance contained in a quartz bucket was sealed up in a glass tube , which was then exhausted and left for an hour .
No loss in weight 1910 .
] Relative Atomic Weights and Sulphur .
95 .could be detected , and was certainly less than 0'000005 grm. ; any loss , therefore , during the half hour required for the preliminary exhaustion must have been quite negligible .
Weighings.\#151 ; The weighings were made on an Oertling 's assay balance carrying a maximum load of 2 grm. Each arm of the beam was divided into 50 parts , and fractions of a milligramme were estimated with a milligramme aluminium rider .
The sensibility of the balance for the loads used was about six pointer-scale divisions per 01 mgrm .
, and an accuracy of five in the sixth place ( 1 in 80,000 on the nitrogen sulphide ) was probably attained .
The pointer oscillations were read with a lens , tenths of a division being .estimated by eye .
In weighing nitrogen sulphide we used the platinum weights alone of the same set with which the measuring cylinder was \#166 ; calibrated .
By using weights of the same set for both operations , any small error due to deviation of the weights from their absolute face values was , of course , eliminated .
The weights were calibrated against each other in air , using another set as counterpoises , according to the method of Richards .
Vacuum corrections were applied to the apparent weights of nitrogen sulphide , and of water or mercury used to calibrate the measuring cylinder , but not to the brass and platinum weights themselves , since absolute values were not required .
Accuracy of Nitrogen Measurement.\#151 ; The calibration of the constant volume has been discussed already .
In order to see whether there was an appreciable change of volume of the measuring cylinder as the internal pressure changed , the exhausted cylinder was filled with mercury to a mark on the capillary .
Air was then admitted into the dead-space , and the movement of the capillary meniscus observed as the internal pressure rose .
It was found that the cylinder expanded by 0-002 c.c. for a rise of pressure of 300 mm. Since in our experiments the pressure of the nitrogen never differed from the atmospheric pressure by more than 100 mm. , this source of error could be neglected .
That the volume of the measuring cylinder was not altered by distortion when filled with mercury was proved by observing the behaviour of the meniscus in the capillary when the whole cylinder was immersed in mercury ; the change , if any , was not greater than 0 0002 c.c. The two thermometers in the manometer bath were previously compared with a standard thermometer at 0 ' , and at 8'*8 , which was about the usual temperature of the water .
In correcting the volume of gas in the dead-space to 0 ' we used the coefficient 1/ 273 ; this was quite accurate enough , as the volume to be corrected was only 1-275 c.c. It was only necessary to know the temperature of the gas in the intermediate capillary to within 5 ' .
Messrs. F. P. Burt and F. L. Usher .
[ May 26 , Accuracy of Pressure Measurement.\#151 ; Since the accuracy of the pressure measurements made in this particular apparatus has already been fully discussed , * it is unnecessary to examine it again here .
We believe that the corrected pressure of the gas was known to within 0-02 mm. , which corresponds with an error of 1 in 28,000 on the N/ S ratio .
The Tensity of Nitrogen.\#151 ; Guye , f discussing the results obtained by Rayleigh and Ramsay , Leduc , and Gray , arrived at the value 1-2507 grm. for the weight of a " normal litre " of nitrogen .
On correction to the latitude of London this becomes 1*2507 x 1-000588 = 1-25144 .
Using the value obtained by Chappuis for the compressibility coefficient between 1 and 2 atmospheres , namely , \#151 ; 0 00043 , we have applied a small correction where the observed pressure differed appreciably from 1 atmosphere .
Atomic Weight Discussion.-\#151 ; The long debated question as to whether the atomic weight of nitrogen was more accurately represented by the number 14-04 or 14-01 has now been generally agreed upon , and the lower value , 14-01 , which has been adopted by the International Commission , may be taken as a very close approximation to the true atomic weight .
The subject has been exhaustively discussed by Guye , } by Gray , S and by Brauner , j| and the balance of evidence seems in favour of a value very slightly lower than 14*01 .
As regards the atomic weight of sulphur , later researches have yielded results not very different from Stas 's value 32*06 .
From the results obtained by Richards and Jonesll on converting silver sulphate into silver chloride , an atomic weight of sulphur can be calculated , but it depends on the value of chlorine , and , unfortunately , a small variation in the value adopted for chlorine involves a large difference in the resulting value for sulphur .
Thus , if Cl = 35-457 , S = 32-070 ; whereas if Cl = 35-460 , S = 32-078 .
The values arrived at by physical methods vary from 32-050 to 32"070 .
Guye , ** by reducing the critical constants of sulphur dioxide , obtained the value 32-065 , taking the density as 2*9266 grm. per " normal litre , " and Baume and Perrot , -j**j* adopting the same procedure with hydrogen sulphide , obtained the value 32-070 , taking the density of this gas as 1*5392 * * * S ** * Gray and Burt , loc. cit. t ' Journ. Chim .
Phys./ 1907 , [ 5 ] , p. 217 .
I 'Bull .
Soc. Chim .
, ' 1905 , [ iii ] , pp. 33\#151 ; 34 , 44 .
S " Uber das Atomgewicht des Stickstoffs , " Inaug .
Dissert .
, Bonn , 1905 .
|| ' Abegg 's Handbuch der anorganischen Chemie , ' vol. 3 , [ iii ] , p. 34 .
IT ' Journ. Amer .
Chem. Soc. , ' vol. 29 , p. 826 .
** 'Comptes Bendus , ' 1905 , vol. 140 , p. 1241 .
++ 'Journ .
de Chim .
Phys. , ' 1908 , vol. 6 , p. 610 .
1910 .
] Relative Atomic Weights of Nitrogen and Sulphur .
97 and the atomic weight of hydrogen as 1-00775 .
D. Berthelot , * calculating the limiting density of sulphur dioxide from the compressibility measurements of Leduc , and of Goye and his colleagues , deduced from the results of the former an atomic weight S = 32-050 , and from those of the latter , S = 32-064 .
There is reason to believe that the form of the isothermal of a readily condensible gas , such as sulphur dioxide , can be accurately determined only by making a large number of different pressure and corresponding volume measurements , and that attempts to extrapolate from two or three points by means of a parabolic formula or by the application of Van der Waal 's equation are not entirely satisfactory We were unable to find any direct check on our N/ S ratio in atomic weight literature , but an instructive comparison was obtained by examining some of the results of Bichards and his colleagues .
From the equations:\#151 ; AgCl/ Ag = 1-32867 , t AgN03/ Ag = 1-57479 , S and 2AgCl/ Ag2S04 = 0-91933 , || we obtained by eliminating silver and chlorine the simple relationship ( S + 64)/ ( 2N + 96 ) = 0-774648 .
By assigning a definite value to either nitrogen or sulphur , the ratio N/ S can be calculated , and this ratio is scarcely affected* by any probable alteration of the atomic weight of nitrogen .
For example , if N= 14-010 , N/ S becomes 0-43683 , whereas if N = 14'009 , the ratio becomes 0"43682 .
The agreement between these numbers and our own ( 0*43687 ) is all the more surprising when one considers the very indirect process by which the former was calculated .
From Bichards ' results sulphur becomes 32-070 if N = 14-009 , whilst our ratio makes S = 32-067 to the same standard .
We suggest that some indirect evidence for an atomic weight of sulphur slightly lower than 32*070 is afforded in the recent paper by Thorpe and Francish on the atomic weight of strontium .
From a determination of the ratios SrCl2/ 2Ag , SrCl2/ 2AgCl , SrBr2/ 2Ag , SrBr2/ 2AgBr , SrCl2/ SrS04 , and SrBr2/ SrS04 , the authors obtain six values for the atomic weight of strontium , namely , 87"638 , 87"646 , 87'645 , 87"653 , 87'661 , and 87"629 , when Ag = 107-880 , Cl = 35-460 , Br = 79-916 , O = 16 , and S = 32-07 .
Now the excellent agreement in the first four values is apparently not quite so well maintained in the last two .
Out of curiosity , the mean atomic weight of * ' Comptes Rendus , ' 1907 , vol. 144 , p. 269 .
t Gray and Burt , loc. cit. , p. 1660 .
+ ' J. Amer .
Chem. Soc. , ' 1905 , vol. 27 , p. 525 .
S Ibid. , 1907 , vol. 29 , p. 826 .
|| Ibid. , 1907 , vol. 29 , p. 844 .
IT ' Boy .
Soc. Proc. , ' A , vol. 83 , p. 287 .
VOL. LXXXV.\#151 ; A. H 98 The Relative Atomic Weights of Nitrogen and Sulphur .
strontium obtained from the first four ratios , namely 87*646 , was inserted in the last two equations , and the corresponding values for sulphur were calculated .
These proved to be 32*068 and 32*066 respectively .
Then , starting with the mean of these two , namely , 32*067 , and keeping the same values as before for the other elements , the atomic weight of strontium was recalculated from the last two equations .
The values obtained were 87*641 and 87*640 , which show an excellent agreement with each other as well as with the mean from the first four ratios , 87*646 .
On the assumption that our ratio N/ S = 0*43687 is correct , we may conclude that if 14*009+ 0*001 be the number assigned to nitrogen the atomic weight of sulphur may be taken as 32*067 + 0*002 .
All the preliminary work in this investigation was carried out at University College , Bristol , and we desire to express our thanks to Dr. Morris W. Travers for his invaluable help and suggestions , and for the continued interest which he took in the research ; to Prof. Francis , who not only was personally responsible for much of the earlier experimental work , but who also afforded us every facility in the prosecution of the research during this period ; and to Mr. O. C. M. Davis , for preparing most of the nitrogen sulphide .
We have also to acknowledge a grant from the Chemical Society , which defrayed the expenses of the earlier part of the work .
The later part , including the analyses with quartz wool , was carried out at University College , London , and we wish to thank Sir William Eamsay for his interest and encouragement .
Finally , we are indebted to Dr. E. W. Gray for his courtesy in placing at our disposal much of the apparatus used .

