June, 1 9 1 3
T H E JOUR.\‘.4L
OF I : V D C S T R I . l L A N D E:‘\‘GI-\-EERISG
Way Association method) of requiring apparatus easily obtainable; of requiring a minimum amount of material for the distillation ; and of requiring a minimum amount of time. All of these are important considerations in a commercial laboratory. TABLEV-EXTTRACTS
O F VARIOUS
SPECIFICATIONS
AfASSACHUSETTS HIGHWAYCOMMISSION’S SPECIFICATIOK FOR REFINED TAR,1909: “It shall contain no body that distills a t a lower temperature than 223 degrees C. : not over 10 per cent by weight shall distill below 270 degrees C . ; and i t shall contain a t least 65 per cent, by weight, of pitch or bituminous material remaining after all bodies up to 360 degrees C. have been distilled.” ILLINOIS HIGHWAYCOMMISSION’S SPECIFICATION FOR TAR BINDER, 1912: “Disti[Zation.-Fractional distillation shall give results within the following limits, all measurements being by volume: “Up to 1 l o o C. the distillate shall not exceed 2 per cent, and shall be free from ammoniacal water. “From llOo C. to 1 7 0 ° C. there shall be not to exceed 5 per cent distillate, of which not more than one-fourth shall be naphthalene. “From 170° C. to 270’ C. there shall be not more than 30 per cent nor less than 20 per cent of distillate, of which not more than one-third shall be solids when cold.” NEW YORK HIGHWAY COMMISSIOS’S S P E C I F I C A T I O N FOR BITUMINOUS MATERIAL“T” (HIGH CARBONTAR), 1912: “ I t shall contain no body that distills a t a lower temperature than l i o n C.; not over 1 per cent shall distill below 235’ C.: not over 10 per cent shall distill below 270° C.; not over 20 per cent shall distill below 300 degrees C. “The specific gravity of the entire distillate shall not be less than 1.02. “The residue from the foregoing distillation shall have a melting point not greater than 7 5 degrees C.” SPECIFICATIONS FOR “BITUMINOUS CONCRETE PAVEMENT,“ ADOPTED BY THE ASSOCIATION FOR STANDARDIZISG PAVXNG SPECIFICATIONS IN JAN., 1912: “ C o d T a r Cement.-So distillate shall be obtained lower than 338’ F., and up to 600 degrees not less than 5 per cent. and not more than 20 per cent of distillate shall be obtained. The distillate shall be of gravity of n o t less than 1.03 a t 60 degrees F. The residue shall have a melting point of not more than 165 degrees F . “In making this distillation an 8 02. glass retort shall be used, and the thermometer suspended so that before applying the heat the bulb of the thermometer is inch above the surface of the liquid.” ROADBOARDO F ENGLAND’S SPECIFICATIONNO. 5 ADOPTED I N APRIL, The t a r shall be free 1911: “SBecification /or T a r No. ,?.--Fractionation: from water, and on distillation shall yield no distillate below 140” Centigrade, nor more than 5 per cent of distillate up to 220 degrees Centigrade, which distillate shall remain clear and free from solid matter (crystals of naphthalene, etc.) when maintained a t a temperature of 30 degrees Centigrade for half an hour. “Between 140 degrees and 300 degrees Centigrade it shall yield not less than 15 per cent. nor more than 2 1 per cent of the weight of the Tar.”
V A L U E IN SPECIFICATIONS
A distillation clause in specifications for road t a r was originally introduced with the idea of governing the consistency. Methods for determining consistency have, however, been so well worked out in recent years that the distillation test is not required for this purpose, and the mistake has often been made of calling for a distillation test not in accordance with the viscosity test. The two are closely related, as is shown by Table IV, showing the total distillate of four refined road tars of differing viscosities, made from the same raw tars, as determined by the American Railway Engineering & Maintenance of Way Association method. The real value of the distillation test is that when properly worded i t defines t o a certain extent the tars from which the product may be produced and the method of distilling them. The results should always be interpreted with reference to the other constants determined. The distillation specifications may be made t o show: I . The detection of water. Water will be shown in
CHEMISTRY
469
the first fraction and may be determined quantitatively. 11. The kind of tar. For this purpose the specific gravity of the oils distilled is indispensable and should .be required. Interpreted with reference to the free carbon and the viscosity, coal t a r and water gas tar may be distinguished by the specific gravity of the oils. 111. The method of manufacture. The detection of a cut back t a r may be inferred from a n abnormal specific gravity of oil for the viscosity, and from a n abnormally high melting point of the residue of distillation. For this purpose the melting point of the residue should always be required. IV. The presence of abnormal amounts of naphthalene. An examination of the Iiquefying point of the oil will indicate abnormal naphthalene. An examination of some of the specifications in use would seem to show little uniformity in the requirements, and few indicate the method to be used, thus invalidating, to a certain extent, the objects aimed a t . I n Table V are shown a number of current specifications calling for approximately the same grade of refined tar. I t will be noted that no two use the same cutting points, that no two specify the same end point and t h a t only two call for the specific gravity of the distillate and the melting point of the residue. The great variety in the specifications takes away no small part of their value, as the direct comparison of the results of different observers is prevented. The importance of the test would seem to warrant its standardization. The important points to be observed in drawing specifications for distillation are: I . Absolute definition of method. 2. Designation of specific gravity of oil. 3. Designation of melting point of residue. 4. Conformance of distillation specification with other parts of the specification. 297 FRANKLIN ST. BOSTON,MASS.
THE DETERMINATION OF SULFATE IN AMMONIUM SULFATE SOLUTION WITH SPECIAL REFERENCE TO THE TESTING OF ILLUMINATING GAS‘ By R. S. MCBRIDEAND E. R . WEAVER
The following paper is a report of a short series of experiments undertaken t o determine the methods most suitable for sulfate determination in the testing of illuminating gas. I t is believed that some of the results obtained may be of general interest. I n the use of the various forms of apparatus for determination of sulfur in gas a solution of ammonium sulfate and ammonium carbonate is obtained, and it is necessary t o determine the sulfate present in such solutions. For official or commercial testing an accuracy of two or three per cent in this determination is ample; i t is, 1 Paper presented a t the Annual Meeting of the American Chemical Society, Milwaukee, March, 1913. This paper reports the same investigation that is given in the latter part of Technologic Paper, No. 20 of the Bureau of Standards. Published by permission of the Director, Bureau of Standards.
470
T H E JOL-R-YAL OF I l Y D F S T R I A L AND E h ' G I N E E R I S G C H E M I S T R Y
therefore, possible t o utilize rapid methods not applicable in problems where great accuracy is needed, The methods applicable here are, in modified form, probably also suited t o use on cement, iron and steel, rubber and other materials where the sulfur determination is important. I n many cases where testing material supplied on contracts requires a large number of sulfate determinations, these methods may be used. I n case of material very close t o the limit of the specification the first quick test can be checked b y more accurate but longer methods; but the larger number of samples would need test only by the short method. The methods for sulfate determination are of three groups: gravimetric, volumetric and turbidimetric. Although most used, the gravimetric procedures are b y far the slowest; in our work two modifications of this group determinations were used, both involving a weighing of barium sulfate. These two are outlined a s methods I and 2 below, but they are of interest only t o show the basis of the comparisons given later. The volumetric processes are in use very little; but for rapid and approximate work in routine testing of a large number of samples of any approximately uniform material they deserve consideration. One volumetric procedure has been used by us. Turbidimetric methods have been little used, but experience with one such method on solutions from the various forms of apparatus t o determine sulfur in illuminating gas has indicated a n accuracy and speed attainable with this method such t h a t i t is promising as a generally useful analytical method. I n the following tests the only substances present in any amount in the ammonium sulfate solutions were ammonium carbonate and very small amounts of silica (dissolved by ammonia from the glass of the apparatus in which condensation took place). The four methods used are as follows: Method I-Gravimetric, with Evaporation to Dryness Dilute or evaporate the sample to 300 cc., add 5 0 cc. of concentrated hydrochloric acid, heat t o boiling, and run in I O cc. of I O per cent solution of BaC1,.2H,O through a tube or burette delivering it a t the rate of 8-10 cc. per minute, stirring constantly during the addition. After evaporation t o dryness on the steam bath, take up with 75-100 cc. hot water, filter, wash till the washings amount t o 2 0 0 cc., ignite and weigh the barium sulfate. The precipitates thus obtained contained a trace of silica, but this could be removed easily when necessary b y moistening with hydrofluoric acid and sulfuric acid, igniting and weighing again. Method 2-Gravimetric, with Precipitation i n Nearly Neutral' Solution Dilute or evaporate the sample of solution t o a bulk of 300 cc., make neutral t o methyl-orange1 by adding hydrochloric acid, and add 2 cc. of the I : I acid in excess. Heat to boiling, add I O cc. barium chloride solution as in method I , boil 5 minutes, let stand on the steam bath a half hour or longer, filter, wash, ignite and weigh the barium sulfate as usual. The 1 Any other indicator sensitive t o ammonia but not affected by carbon dioxide, may be used.
1701. 5 , NO. 6
conditions of precipitation are made uniform and the loss by solubility of barium sulfate is negligible for the present work. Method 3-Volu.unetric This method as used by us is more like that of Holligerr than any of the others previously recommended. Make u p a solution of 50.0 g. BaC1,.2H,O per liter, and one of 30.5 g. K,Cr,O, per liter. To 300 cc. of water add I O cc. of I : I hydrochloric acid and exactly 15 cc. (from pipettes) of each of the above solutions; heat t o boiling, add two or three drops of a dilute solution of ferric chloride, then slowly add I O per cent ammonium hydroxide until a precipitate forms which does not redissolve on stirring. Add 5 cc. of the ammonia solution in excess, boil 5 minutes, filter and wash thoroughly with hot water. Allow the filtrate t o cool, acidify with hydrochloric acid, add two grams of solid potassium iodide or a solution containing that amount, and titrate with tenth-normal sodium thiosulfate, using starch as a n indicator. The amount of thiosulfate solution used in this blank is t o be subtracted from the amount used in each subsequent determination. To make a determination of the sulfate in a solution, dilute or concentrate the sample taken t o 300 cc. ; add I O cc. of I : I hydrochloric acid; heat t o boiling; add 15 cc. of barium chloride solution from the pipette used in the blank test; boil five minutes; add 15 cc. of potassium bichromate solution and a few drops of ferric chloride, precipitate with ammonium hydroxide and complete the determination exactly as in the blank test. From the amount of thiosulfate solution used subtract the amount required in the blank. The remainder represents the sulfate in the, sample taken. The reactions upon which the method is based are:
+ +
H,SO, BaC1, BaC1, K,CrO, 16HCl 6KI 2K,CrO, 8H,O 21 2Na,S,O,
+
+
+
+
=
BaSO, 2HCl BaCrO, 2KCI 61 IoKCl 2CrClg
=
Na,S,O,
= =
+
+
+
+
+ 2NaI
From the above equations i t will be seen that three Na,S,O, are equivalent to one S, hence I cc. N / I O Na,S,O, solution is equivalent to 1.069 mg. (0.0165 grains) of sulfur. The above method was selected in preference t o other methods previously used, which have been open t o the following objections: I . I n those using a hydrochloric acid solution of barium chromate the reducing action of the acid on the chromate has introduced serious errorsa 2 . I n those which use exactly equivalent amounts of barium chloride and potassium chromate 'it is somewhat difficult to make the solutions of exactly the right strength and a blank test must be made.3 3. The use of a suspension of barium chromate or of Z . anal. Chem.. 49, 84 (1910). Andrews, A m . Chem. J . . 11. 567 (1889); Pennock, J . A m . Chem. SOC..26, 1265 (1903); Holiiger, Z . anal. Chem., 49, 84 (1910); Bradley, Chem. Env., 18, 26 (1911); Reuter, Chem.-Ztv., 357 (1898). 3 Roemer. Z . anal. Chem.. 49, 490 (1910); Precht, 2. anal. Chem., 18, 1
2
521 (1879).
the solid is less convenient than that of solutions and as i t is sometimes difficult t o obtain the barium salt free from alkali chromate, a blank test should also be made with this method.' 4. I n all methods the solubility of the barium chromate should be taken into account.
All of these sources of error or difficulty are taken care of by a single blank test when the chromate solution is slightly stronger than the barium chloride solution, as in the method described. The solubility of barium chromate is somewhat affected by the concentration of FIG.I-TURBIDIYETER ammonium chloride and of ammonium hydroxide in the solut i o n ; but by having the same bulk of solution, adding the same amount of hydrochloric acid and precipitating in the same manner in the blank as in the
Milligrams
ide. as recommended by Holliger,' aids greatly in collecting the precipitate of BaCrO, and BaSO, for the filtration. Method 4-Tzirbidimetric2 The turbidimeter used (shown in Fig. I ) consists of a cylindrical glass tube graduated from the bottom in centimeters, surrounded and held in place above the light by a brass tube. The light used in most cases was a 16 candle power carbon filament lamp. so placed t h a t a straight portion of the filament extended diametrically across just below the bottom of the tube. The following method was adopted after tests, the results of which are given later in this paper; i t is recommended for general use when a turbidimetric procedure is wanted. The directions are given for use in connection with a n apparatus for sulfur in gas, b u t can easily be applied t o solutions otherwise obtained. The condensate and washings from the sulfur apparatus are neutralized with hydrochloric acid,^ then 2 cc. of the I : I acid added in excess. The solution is measured (to the nearest cubic centimeter) and a 90 cc. portion placed in a small beaker for the test.
per
of &u/+Au7-
/OQc.c. of cJb/zLf~~um
FIG.2
determination, the effect of this source of error is also eliminated. The use of a small amount of ferric chlor-
It may be measured in a n ordinary graduated cylinder 1 LOG. Cil.
Hinds, J . A m . Chem. S o c . , 18.661 (1896);22, 269 (1900),24, 848 (1902).
I
472
T H E J O U R X A L OF I L V D U S T R I A L AA'D EI\-GINEERI/VG
if it is certain t h a t the ratio of the small cylinder t o the larger measuring vessel first used, is correct. While the solution is between 2 5 and 30 O C., I O cc. of a I O per
Fig. 3
CHEJfISTRY
Vol. 5 3 NO. 6
(Fig. 2 ) corresponding t o the depth of a liquid in the turbidimeter measurement is multiplied b y the number of cc. of neutralized condensate and divided by 90 (the cc. of sample taken) t o give the total amount of sulfur obtained by burning the gas. For convenience in this calculation the condensate after neutralization can be made up t o 2 7 0 or 360 cc. and the observed values for the 90 cc. portion are then multiplied by 3 or 4. The curve of Fig. 2 may be used b y any observer if care is exercised in following exactly the method given in the preceding paragraph: the acidity and temperature limits must be particularly observed ; the character of light used is less important. I n any case it is well for a n inexperienced observer t o take k n o w n amounts of a standard sulfuric acid solution and run through the test until assured that his observations are consistent and correct. Since little was known of the effect of different conditions of operation on t h e result of the tests, a number of special comparisons were made, t h e results of which are given in the following paragraphs. The incandescent filament gives a sharper end point
cent solution of BaCl,.zH,O is added and the whole stirred vigorously until the precipitation appears t o be complete. (This usually requires less than one minute's stirring). The suspension is now poured into the clean and dry turbidimeter tube, a small portion at a time, until the filament of the lighted lamp disappears from view. After the first trial the solution is poured back into the beaker, stirred vigorously and the observation of the point at which the filament disappears is repeated until this is fixed within one millimeter. During the time the solution is being added the portion remaining in the beaker should be kept well stirred so t h a t the small portions added from it will be representative of the whole bulk. The first time the solution is poured into the observation tube the latter must be dry and clean; in cleaning it care should be taken t o remove, without scratching the tube, the film of sulfate which adheres so persistently . t o the bottom. MW*rnnrs o f Sulpkur Fi-/O&C. c. ofsdu/Lun. Further tests on a secFig. 4 ond 90 cc. of the original neutralized solution of than any other source of light so far devised;I it is 1 See Jackson, J . A m . Chem. S o c . , 23, 7 9 9 ( 1 9 0 1 ) ; Parr, THIS JOUR'sulfate should be made, if considerable accuracy is NAL, 1, 689 ( 1 9 0 9 ) ; Leighton, U. S . Geoloukal Survey. Wafer SufiP6 needed. The amount of sulfur indicated by the curve and Irrigation, PaPer 161, 27 (1905).
June, 1 9 1 3
T H E J O U R - T A L OF I , T D C S T R I A L AlVD EIVGIIVEERING C H E M I S T R Y
not affected by draughts, does not smoke nor crack the tube, and can be conveniently placed under a table or in any other out of the way place. A series of comparisons using the incandescent filament, the standard candle and the 0.j mm. slot recommended by Leightonr is given in Table I . I n each case the same liquid was used with each source of light in turn. I t can be seen t h a t the three light sources give excellently agreeing results .
Fig. 5
The effect of using different precipitants as proposed b y various persons using the turbidimeter for sulfate determinations, is shown in Fig. 3. I t will be observed t h a t there is less variation where I O per cent BaCl,.zH,O solution is used than with other precipitants;
473
Fig. 4 shows the effect of small amounts of hydrochloric acid upon the turbidimeter curve. Larger amounts caused much greater variations. Fig. 5 shows the effect of temperature of precipitation upon solutions of average strength. Fig. z is the curve drawn from the results obtained under the conditions of precipitation recommended above. To show the accuracy of the method as used, the tests of one series from which our conclusions are drawn have been summarized in Tables I1 and 111. This summary shows the average and the maximum deviation of individual tests from the value calculated from the amount of standard sulfuric acid used, i. e . , the errors which were found by making a single precipitation of barium sulfate and suitable observations on it. When more than one portion is precipitated and used for a determination the accuracy would be materially increased, so t h a t in tests on three portions of a solution the average of the three ought not t o be more than one per cent in error. I t should be remembered t h a t three tests can be made in this way much more quickly than a single test by any other method.
TABLEI1 Average error Maximum error No. of deterSulfur (mg. per 100 cc.) (per cent) (per cent) minations 1-2 2.9 8.7 13 2-3 3.5 7.1 13 3-4 2.5 i.7 17 TABLE I-COMPARISON O F RESULTSWITH DIFFERENTLIGHTS I S TUR4-5 2.6 6.3 8 BIDIMETER 5-6 1.3 5.2 7 Incandescent Electric light belolv a 6-8 2.6 5.6 7 filament Candle 0.5 mm. slot 8-11 3.4 6.3 5 c c---m--m ___ Depth of Milligrams Depth of Milligrams Depth of Milligrams 2.7 8.7 70 All values solution of sulfur solution of sulfur solution of sulfur 1.12 1.12 25.0 1.13 25 . O 24.0 From its very nature, as a procedure using a stand20.4 1 22 20.6 1.21 20.3 1.22 ard of similar nature as basis of calibration, the tur1.79 1.77 12.2 1.77 12.3 12 3 11.6 1.86 11.8 1 .s5 bidimetric method must give, on the average, correct 11.6 1.86 2.51 2.48 8.4 8.5 8.5 2.48 results; the variation of individual results from the 2.93 3.01 7.2 2.93 7.O 7 .2 3.60 3 50 5.9 6.1 5.9 3.60 TABLE111-SHOWING PERCENTAGE OF ERROROF DETERMINATIONS BB4.09 3.95 5.2 5.4 5.4 3.95 TWEEN 2 AND 8 MG. PER 100 cc. 3 9 5.55 4.1 5.25 4.1 5.25 Number of determinations Error (per cent) 3.7 5.93 3.8 5.77 3.7 5.93 0.0-0,5 11 6.70 6.26 3.3 3.4 6.50 3.5 0.5-1.5 12 1.5-2.5 5 the addition of oxalic acid gave a finer precipitate and 2.5-3.5 7 a sharper end point, but successive determinations were 3.5-4.5 3 more discordant. Agreement of duplicates and cer4.5-5.5 5 5.5-6.5 6 tainty of result is most desirable in this method, since 6.5-7.5 2 in any case the results are fixed by comparison with 7.7 1 \
results on known amounts of sulfur. Variation of a single test from the average of all is therefore serious, but the shape of the calibration curves is not of influence, so long as regular and calibration tests are made in the same way. Loc. cit.
52
average is, therefore, the important consideration. The data of Tables I1 and I11 show that when used with only ordinary care it is applicable t o work where approximate values (to 4 or j per cent of the amount
THE JOL-RA\-AL OF ISDL-STRIAL A-YD E-YGIXEERISG CHEMISTRY
474
of sulfur present) are sufficient. To give a n idea of the results to be had b y the other three methods, Table IV has been included. These data show the results which were obtained under conditions distinctly unfavorable for accurate results, only 6 to 30 mg. of sulfur and large and varying amounts of ammonium carbonate being present in the samples used. These conditions were chosen as representing those with which we had t o deal in the gas testing; and if in other work larger amounts of sulfur were present or greater care was exercised no doubt much more exact results could be had. For gas testing the accuracy here shown is ample, since i t is greater than the accuracy obtainable in the condensation of the sulfate.
Vol. 5 No. 6
fuels can be classified in four groups, the general principles o f which are as follows: ( I ) The gas is burned and the sulfur oxides in the products of combustion are condensed or absorbed, oxidized to sulfate, and determined as such. ( 2 ) The sulfur compounds are oxidized by liquid reagents giving the sulfate, which can be determined directly. (3) The sulfur compounds are oxidized by burning the gas but the sulfur is absorbed and determined as sulfur dioxide. (4) The sulfur compounds are reduced to give hydrogen sulfide, which is then absorbed and determined. TABLE IV-COMPARISON OF METHODS FOR SULFATEDETERMINATION. The methods of Group I have been most used in IN STANDARD HzS04 SOLUTJON SULFUR(MG.PER G . OF SOLUTION) American practice and are generally considered the DETERMINED B Y DIFFERENT METHODS IN PRESENCE OF EXCESSOF .4MMONIUM CARBONATE EXCEPT AS most accurate and convenient, either for works conOTHERWISE NOTED trol or official testing. Tests have been made on five Method 3. Volumetric forms of apparatus of this group and one or more Volumetric ___7 against Oxalate standforms of each of the other classes; but the principal h*aOH and Method 1 Method 2 ard (no carHzE04 investigation was limited to the three forms, all of benzoic acid Gravimetric Gravimetric bonate added) standard group one, which appeared t o be most satisfactory for 1.589 1.591 1.597 1.602 1.595 93 91 93 1.597 1 ,602 general use. These are known, respectively as the 90 90 92 1.600 1.597 Referees apparatus, the Hinman-Jenkins apparatus, 92 1.609(a) 99 1.599 1.601 and the Elliott apparatus. 92 1.596 97 1.602 1.596 ... 1.589 92 96 92 The Referees apparatus is too well known to re... 1.574 89 88 99 quire a detailed description here. I t consists of a ... 1.575 90 92 91 .. 94 1,600 ... ... Bunsen burner surrounded b y pieces of crystallized __ __ __ ammonium carbonate, from which the products of Av. 1.591 1.592 1.595 1 ,600 1.589 combustion of the gas, mingled with vaporized am( a ) Omitted from average. monium carbonate, pass into a condensing cylinder The first set of results in Table IV by Method 3 are where the sulfur oxides are absorbed in the water calculated on the basis of the value of the thiosulfate solution obtained b y titration against a permanganate condensed from the products of combustion and are solution, whose value was obtained against sodium oxidized, in the presence of the ammonia, to amoxalate. I n the second series the thiosulfate was monium sulfate. The Elliott apparatus is essentially a large Referees standardized against the sulfuric acid value given in apparatus, the principal difference being the use of the first column of the table. For accurate work the gravimetric methods, with the two condensing cylinders through which the gases precautions which have been carefully developed by successively pass. The Hinman-Jenkins apparatus' many experimenters, are to be preferred; but for differs from the Referees greatly in form, but in rapid work the turbidimetric and volumetric methods principle only in the use of concentrated ammonium are very useful. They can be applied in many lines hydroxide instead of ammonium carbonate as a of work other than gas testing, since in each the value source of ammonia. of the standard is fixed against similar material whose To compare the several forms of apparatus, two or value has been previously established by other methods.' more were operated a t the same time, being supplied If the calibration and testing procedures are identical with gas from a common source. Suitable prethe variation of individual testsfrom the average or cautions t o maintain a uniform supply to each apcorrect value is the important factor; this variation paratus and t o prevent errors due t o sulfur in the air has been shown by our experience t o be about one of the room were taken. The meters used were freper cent or less in the volumetric and five per cent or quently compared to determine their relative accuracy. less in the turbidimetric method. One series of comparisons made upon the three BUREAUO F STANDARDS apparatus to determine the best conditions of operaWASHINGTON tion and sources of error is summarized in Table I. _____~______ T H E DETERMINATION OF SULFUR IN ILLUMINATING GAS' The results tabulated include only those obtained under normal working conditions, and not those obtained B y R. S. MCBRIDEAND E. R. WEAVER *when variations were made which would be avoided The methods for determination of sulfur in gaseous in practice. The sulfate in the condensed liquids was 1 Paper presented at the Annual Meeting of the American Chemical determined as barium sulfate by a method closely Society, Milwaukee, March, 1913. Published by permission of the Director of the Bureau of Standards. This paper is a report on tests made at this resembling t h a t of Johnston and Adams,' the weight ~
-
~
Bureau, preparatory to the recommendation of a method for official inspection of gas; a more complete report on this same investigation is to appear soon i n Technologic Paper No. 20 of the Bureau.
1
J. Ant. Chem. Soc , 28,
2
Ibtd., 33. 829-45 (1911).
543 (1906).