1264
INDUSTRIAL AND ENGINEERI-VG CHEMISTRY
Vol. 19, No . l l
6-A composition of CaO 67, A1203 10, and Si02 23 per cent It is not the aim of the paper to show that ferric oxide, requires, under the conditions of this study, a temPeratVe of magnesia, or the alkali oddes are desirable or undesirable 1500' C. to effect complete combination. 7 - 4 composition of CaO 63, MgO 3, NazO 1, +?037, pezo8 in Portland cement except in that they lower the temperature 3, sioz23 per cent, which approaches the composition of com- a t which a certain base composition can be burned to commercial Portland cement, requires a temperature of about plete combination. This base composition (CaO 67, &OI 1325 c. for complete combination under the conditions of these 10,s i o z 23 per cent) approaches the average composition of experiments. This is a decrease of 175" C. below the temperaPortland cement when the several components are referred to ture required for the base composition. 8-The common belief that any material that will cause an the oxides of lime, alumina, and silica. Furthermore, these increase in the amount of liquid formed in such a system, at a data indicate nothing a t all with respect to any other base given temperature, will likewise cause an increase in the combina- composition, and it is not permissible to assume similar tion of lime is true in this composition for magnesia and ferric oxide, but to a very limited extent for soda or potash, The effects for other compositions. The entire field of Portland charges containing these latter oxides were distinctly more fused cement is now being explored and the data are being correlated than the ones without them, but no great increase in combination with the properties of the products in service. was effected by their presence. A possible explanation is given.
Influence of Neutral Salts on the Plumping of Hides' By R. 0. Page and A. W. Page ~ ' O o L s T O ST A Y N E R I E s , R'OOLSTON,
Data showing the influence on the plumping of cowhide of the chloride and sulfate of sodium and of calcium chloride in concentrations up to 4 normal, and a t the four pH values of 2,5,8, and 11, are given in detail. The pH of the solutions employed greatly affects the result of the addition of neutral salts. I n acid solutions, addition of the three salts studied reduces the plumping t o a minimum value of 0.8 a t a concentration of 0.75 normal. A t greater concentrations t h a n this t h e plumping increases again, this increase being most marked with calcium chloride, sodium sulfate giving results similar to those given by sodium chloride. A t t h e pH values 5 and 8, the three salts give widely differing results. Sodium sulfate gives in both cases a rather flat plumping curve with .a minimum a t a concentration of 0.75 normal, while sodium chloride gives a maximum a t about t h e same concentration, this maximum being less pronounced a t pH 8 than a t pH 5. Increasing concentration of calcium chloride increases the plumping a t both these pH values, rapidly a t first, then more slowly a t concentrations between 1 and 3 normal, and finally more rapidly u p t o 4 normal. The plumping is more marked a t pH 8 t h a n a t pH 5. Hide pieces plumped in the more
N E W ZEALAND
concentrated solutions lose calcium chloride rapidly and completely on washing, but lose their plumping much more slowly, hide plumped in a 4-normal solution retaining its plumping even after soaking for a week in distilled water. In alkaline solutions increasing concentration of sodium chloride depresses t h e plumping, but much less t h a n in acid solution. The least plumping (1.27) occurs a t t h e maximum concentration studied. Sodium sulfate, on t h e other hand, depresses the plumping much more rapidly to a minimum of 0.84 a t 1.5-normal concentration, the plumping increasing again slightly in more concentrated solutions. Hide plumped in sodium hydroxide solutions of pH 12.5 reaches apparent equilibrium a t the end of a week, but in a calcium hydroxide solution of the same pH the plumping increases in t h e second week. I n neither case does the hide fall t o its initial thickness on restoring t o its original pH value, but on bating in both cases it falls further but not to its original substance. There appears to be a close connection between the nitrogen dissolved from cowhide by solutions of neutral salts and the plumping in such solutions.
.............. ECENT researches have indicated that as far as collagen and gelatin are concerned specific-ion effects, corresponding to the Hofmeister series, have real significance and cannot always be attributed, as LoebZhas done, to error of experimental method arising from neglect to control the p H values of the salt solutions employed. Stiasny* and his collaborators found that, in regard to the influence of salts upon both the dispersity of gelatin and the action of enzymes on hide substance, while the Donnan equilibrium is not applicable, the anion action is in accordance with the Hofmeister series. Ostwald, Kuhn, and Bohme4 found that the swelling of gelatin a t constant p H reveals the
R
1 Received June 16, 1927. Presented before the Division of Leather and Gelatin Chemistry at the 74th Meeting of the American Chemical Society, Detroit, Mich., September 5 to 10, 1927. 9 "Proteins and the Theory of Colloidal Behavior," 1st ed., Chap. V. 8 Collegium, 1S26, 13, 23, 67. 1 Kolloidchem. Beihefte, PO, 412 (1926).
existence of a specific-ion series; while Thomas and Fosters showed that in the destructive action of neutral salts on hide powder over a narrow pH range around the isoelectric point of collagen, once more a specific-ion effect is apparent corresponding to the Hofmeister series. Gustavson6 reviewed the literature, a t the same time suggesting an explanation of the action in question, and later' showed the important influence of previous treatment with neutral salts on the behavior of hide powder during tanning. McLaughlin and Theis,8 on the other hand, while confirming the finding of Thomas and Foster that sodium chloride dissolves hide powder more readily than does sodium sulfate, discovered that in the case of untreated hide the sulfate dissolves as much as the chloride. T m s JOURNAL, 17, 1162 (1925). e J . A m . Leather Chem. Assocn., 21, 206 (1926). 7 I b i d . , 21, 366 (1926). 8
a Cdtegium, 481 (1926).
November, 1927
I-VDUSTRIA L AND ENGINEERI.liG CHEMISTRY
1265
could be obtained with pieces from the butts of different hides as long as their substance was fairly uniform. As pointed out by Porter,'O however, differences of thickness alter greatly the degree of plumping, which is roughly inversely proportional to the initial thickness. By taking average values satisfactorily smooth curves were obtained, curves which, in the case of sodium chloride and sodium sulfate a t p H 11 and sodium chloride a t p H 5 , were duplicated almost exactly in check determinations a t a considerably later date. As was to be expected, however, it ivas found that the averaged results, though closely parallel to those previou4y obtained, were higher or lower according as the average inExperimental itial substance of the hide pieces mas lower In determining the degree of plumping, the method of Kilson and Gallung n-as used. Pieces of unhaired cou-hide or higher. T h e salts sodium Figure 1-Plumping a t pH 2 about 1 inch (2.5 cm.) square were taken. To insure, as far chloride, calcium chloas possible, uniformity of structure and absence of excessive variations in thickness, they were in all cases cut from the ride, and sodium sulfate were chosen for study with the idea same portion of the butt. They were bated a t a p H value of discovering any differences of behavior that exist both of 7.9, washed in cold running tap water for 24 hours, and between chloride and sulfate anions and between sodium and calcium cations their thicknesses in this standard condition derermined by means of a gage pressing on the hide a t constant pressure Results-Specific Action of Salts on Plumping of Hide for 2 minutes. The pieces were than placed in 260 cc. of AT PH 2-Figure 1 shows the action of various concensalt solutions of different known concentrations a t the required pH. These solutions were not buffered, the correct trations of sodium chloride, sodium sulfate, and calcium pH being obtained by the addition of acid or alkali. The chloride, respectively, on the plumping of hide a t a pH value added acid was hydrochloric in the case of the chlorides and of 2. The effect of sodium chloride was investigated for sulfuric in the case of the sulfates, while sodium hydroxide solutions acidified with both sulfuric and hydrochloric acid. N o t e T h e figure for degree of plumping is the ratio of the final to was used for experiments a t pH 11. In each case preliminary determinations were made t o find what additions of acid initial gage reading. All three salts give curves rapidly descending to a minimum or alkali were needed to keep the pH approximately cona t a concentration of 0.75 equivalent per liter of added stant. After 24 hours the salt solutions were changed and the salt. Then with increasing concentrations all the curves hide pieces kept in the fresh solutions a t the required pH unexpectedly rise. With sodium chloride the rise is uneven, for a further 48 hours. The p H values a t 2 and 11 were b e i n g m u c h m o r e determined electrometrically; a t 5 and 8 colorimetric methods r a p i d a t first (up(- ' ' were used, the final p H value being checked electrometrically. to about 1.3 equiva- yi4 In some instances it was impossible to keep the p H con- lents) than at higher 2 stant over any considerable time. This was especially the concentrations; it is a 12 case a t a p H of 5, mhen there was a marked tendency for more or less uniform the hydrogen-ion concentration to fall, apparent equilibrium throughout for the ki0 being still not reached a t the end of 10 days. As trial ex- other two salts. and periments showed that in the presence of salt solutions of considerably greater ob the strengths used a small change of p H had little or no for calcium chloride 025 OS Pl5 I I5 2 15 3 effect on the plumping, the method adopted as the one most than for either the Eguinrlmtr of b l f p r r h b chloride or the sulfate likely to give correct results was to keep the hide pieces for Figure 2-Plumping at pH 5 a uniform time in salt solutions whose acidity or alkalinity of sodium. had been adjusted in accordance with previous determinaThe portions of the curves representing concentrations tions, and then to adjust the pH again a few hours before below 0.75 normal are similar for sodium and calcium chlothe final gage readings were taken. rides, but in the case of sodium sulfate the reduction of plumpThe temperature of the chloride solutions was maintained ing does not increase so rapidly with increasing concentration a t 15' C. I n the case of the sulfate a temperature of 20' C. or reach so low a turning value. was employed to render possible the use of solutions up to The two curves for sodium chloride show fairly close 2.6 normal concentration. correspondence apart from the very much greater initial In order to even up deviations caused by differences in plumping in one due to the plumping power of hydrochloric the hide structure and thickness of the pieces, measurements acid being almost double that of sulfuric. The slight increase were made on six pieces of hide for each determination, and of plumping caused by the addition of 0.25 equivalent of their average taken. It was found that very close agreement sodium chloride to the dilute sulfuric acid solution is ob-
In actual tannery practice the influence of salts over wide ranges of concentration and p H values is of great importance. Ordinary wet-salted hides contain an approximately saturated solution of sodium chloride; in the soaks there is up to 3 per cent of sodium chloride in a solution of pH 5 to 6; in all those liming processes in which calcium chloride is added to reduce the sodium hydroxide content of the sodium sulfide present, the hide is acted on by sodium chloride solutions a t a p H of not less than 12.5. In bating. ammonium chloride and calcium chloride are present a t a p H of 8; in the pickling process the pickle liquor is a solution of sodium chloride up to 3-molar strength a t a pH of 1 to 1.5; while chrome tanning itself takes place in the presence of varying quantities of neutral salts at a pH of from 3 to 4. It was therefore expected that research in regard to the influence of neutral salts on the plumping of cowhide over \vide ranges of p H and of concentration 1%-ouldgive not only information concerning the action of those salts on unaltered collagen, but also more exact data as to the phvsical condition of the hide in the various tannery proce
*
I
9
THISJOURNAL, 16, 376 (1923).
10
J . A m . Leather Chem. .4srocn., 20, 282 (1925).
1266
INDUSTRIAL A-VD ENGINEERING CHEMISTRY
Vol. 19, No. 11
Applicability of Donnan Equilibrium
It was noted above that certain of the results obtained for the action of neutral salts on the plumping of hide in acid solution agree with those calculated from the Donnan membrane equilibrium. For monovalent ions we have the equation'2 e = -2(x
+ u ) + 4 4 ( x + u)' +
22
where e is the excess of concentration of diffusible ions of the hide over that of the external solution, and is proportional to the plumping; u is the concentration of the cation of the added monovalent salt; and z is the concentration of the positive collagen ions in the hide. Rewriting and expanding, we have, when x u is greater than z (which is true a t pH 2 when the concentration of added salt is much greater than 0.25 normal), the approximate result
+
18'
QJb
s f
14
%
$,.: 10
by McLaughlin and Theis" in their work on t h e s o a k i n g of salted hides, in which they measured the degree of plumping by the change of weight, not of thickness. Sodium sulfate, on the other hand, gives almost t h e reverse
I n the case of sodium chloride in a hydrochloric acid solution of pH 2, we have 2 = 0.01. We may look upon u as the concentration of the added salt. Accepting Wilson's figure of 750 for the equivalent weight of collagen, and assuming that the hide contains 80 per cent of water, we may take as an approximate value z = 0.25. Then, assuming that z is constant, we obtain the following values for e: u = 0 0.25 e = 0.23 0.06
0.50
0.75
1.0
2.0
0.03 0.02 0.016 0.008
These points when plotted give for values of u up to 0.75 a curve similar in shape to that obtained experimentally for sodium chloride in hydrochloric acid solutions. For values of u higher than 0.75 they give a curve asymptotic to the x axis (Figure 5). Actually z is not constant, but is inversely proportional to the plumping. Variation in z due to this cause cannot, however, produce a minimum, but only a slight flattening of the curve shown in Figure 5. If equation (1) is to be applicable to solutions containing concentrations of salt greater than 0.75 norlb mal, then a t these concentrations z2/u must increase with $'6 € increase of u. $ I4 Possibly the combination of the added chloride with the col- &'' lagen by means of secondary valencies od may be equivalent to such an increase of Z S J c o l l a g e n ions, i n Lq,l/mts ofso/tprr/,tcr which case the rise of Figure &Plumping at pH 11 the plumping curve would be proportionate to the extent of such action a t any given concentration. This is in agreement with the fact that calcium chloride gives a plumping curve with a steeper rise from the minimum than the chloride or sulfate of sodium. but does not explain the close correspondence between the action of these two salts. On the other hand, it is possible that the action of the neutral salts may so affect the cohesive forces in the hide that in the relation e = cv, c is no longer a constant. If c decreases, then we may have plumping without any increase in e. Probably both factors play a part in the plumping action of neutral salts.
''
12
Wilson, "Chemistry of Leather Manufacture," p. 107.
November, 1927
INDUSTRIAL A N D ENGINEERING CHEMISTRY Solution of Nitrogen
The amounts of dissolved nitrogen in the final solutions which had been in contact with the hide pieces for 48 hours were determined to see if they bore anyrelation to theplumping. The sodium chloride solutions contained very little dissolved nitrogen, the maximum, calculated as hide substance, being 0.18 per cent, and the minimum 0.10 per cent, of the original net weight of the hide pieces. There was little variation with change in concentration; though a t p H values of 5 and 8 a slight maximum was shown a t a concentration of 0.75 normal, while a t pH 11 there was a steady reduction from 0.18 per cent in the absence of salt to 0.10 per cent in 4-normal concentration. Sodium sulfate a t pH 2 and 11 shoved practically no dissolved hide substance, but a t p H 5 and 8 the amounts dissolved were comparable with those obtained for sodium chloride-results agreeing with the work of hlcLaughlin and Theis.8 Calcium chloride a t p H 2 dissolved only slightly more than the sodium salt, but a t p H values 5 and 8 the dissolved nitrogen increased markedly with the concentration, a t pH 8 rising abruptly t o a maximum of 1.5 per cent in 4-normal solution. The dissolved nitrogen thus corresponds roughly to the observed plumping, suggesting that the same factors are responsible for both phenomena. It may be inferred, then, that the observed variations in the plurnping of hide in sodium chloride, sodium sulfate, and calcium chloride solutions u are best explained as Figure 5-Agreement of Experimental due to specific ionic a n d Theoretical Values action of the secondary valencies which combine the smaller primary valence constituents of the collagen molecule. Evidence as to the Nature of the Plumping Action of Salts With the view of throwing further light on the nature of the plumping produced by concentrated salt solutions, pieces of hide, after being kept a given time in calcium chloride solutions of 2- and 3-normal strength, a t p H values of both 5 and 8, were placed in successive changes of distilled water. Gage readings of the pieces were taken a t each stage and the calcium chloride contents of the changes of water determined. After the final washing, the amounts of calcium chloride left in the hide pieces were obtained by ashing. It was found that in all cases the salt was removed rapidly and completely; but the hide from 4 N calcium chloride solution retained almost its full plumping, even after the whole of the calcium chloride had been removed. Hide pieces kept in the salt solutions of lower concentration than 4 normal for from 2 to 6 days gradually lost their plumping, the rate of fall being, roughly, inversely proportional to the length of time during which the calcium chloride solutions had acted. If allowed to act for a sufficient length of time, %normal and 3-normal solutions are, like still stronger ones, able to produce permanent plumping. The results a t p H 5 are summarized in Table I. The results a t p H 8 were quite similar.
Table I-Degree SOAKINQ PERIOD -2 Days 2 N
1267
of P l u m p i n g after Various L e n g t h s of T i m e i n C a l c i u m Chloride Solution (pH 5) DAYS3N
4 N 1.42 1.48 1.47 1.45 1.45
4 DAYS 3 N 1.34 1.11
7 DAYS 2 N
1.44
12 DAYS
2 N 1.43 1.34
1.18 1.10
It is evident that in concentrated solutions calcium chloride causes a definite permanent alteration of the collagen itself, a finding that is in complete agreement with Gustavson’s work on the behavior of neutral salt-treated hide powder towards tanning agents. It would appear that the change postulated by Gustavson in the aggregation of collagen which has been acted on by neutral salts is accompanied by a change in the physical condition of the treated hide. Quite possibly in the liming process, more especially in the longer methods, the plumping is not due wholly to the p H value? part being caused by the specific effect of calcium ions. If this is so, it would explain the importance attributed to the plumping of the hides in sole-leather tanning-for this plumping would, in part, be a measure of the increased power of tanning fixation of the collagen. In an attempt t o throw more light upon the plumping of hide in strongly alkaline solutions, and in particular to distinguish between effects due to p H and those due to specific-ion action, hide pieces were kept for periods of 1 and 2 weeks, respectively, some in a saturated lime solution (with excess of lime present) and the others in a sodium hydroxide solution maintained a t the same pH (12.5) as that of the lime solution. The plumping was determined a t the end of the respective periods, after which the pieces were placed for 24 hours in a normal sodium chloride solution kept a t a p H of 2 by the addition of hydrochloric acid. They were then again gaged, neutralized with cold saturated sodium bicarbonate, washed, and finally kept in sodium phosphate buffer solution for 48 hours, after which time the plumping was measured once more. This method was found t o give almost the same results as were obtained by placing the swollen hide pieces directly into a buffer solution a t p H 8, and was adopted as it was considered that the complete removal of the lime would make the results of the experiments with calcium hydroxide and sodium hydroxide more strictly comparable. These results are shown in Table 11. Table 11-Degree
of P l u m p i n g (Final) in Calcium Hydroxide a n d in
S o d i u m Hydroxide CALCIUM HYDROXIDE SODIUM HYDROXIDE TREATMENT 1 week 2weeks 1 week 2 weeks 1.47 1.47 Original solutions, pH 12.5 1.33 1.35 N NaCI, pH 2 1.03 1.09 1.10 1.11 Buffer, pH 8 1.23 1.27 1.18 1.24 Bate 1.03 1.10 1.03 1.09
It will be seen that, whereas in sodium hydroxide solution the plumping reached equilibrium a t the end of a week, with calcium hydroxide the plumping was greater after 2 weeks than after 1, an increase which, in the light of the previous results, is probably due to the specific action of the calcium ion. Further, in neither case does the hide fall in the normal sodium chloride solution a t p H 2 as much as would be expected, nor does it fall to its initial thickness on being brought back to a p H of 8. It is noteworthy that, while the pieces treated with calcium hydroxide show greater plumping at a pH of 8 than those treated with sodium hydroxide, in salt solution at p H 2 the position is reversed. In both cases an increase of time in the alkaline solution decreases the amount of falling. After the first experiments it was thought that the difficulty in effecting complete reduction of plumping might be due
INDUSTRIAL AND ENGINEERING CHEMISTRY
1268
to the formation in alkaline solution of the second form of ~ o l l a g e n 'and ~ was in accord with the work of Miss Lloyd'4 on the action of alkaline solutions on gelatin. It was for this reason that in the final work the hide pieces were brought to a p H of 2, as from the results of Wilson and KernI6 such treatment should restore the collagen to its initial condition. It proved, however, to have little influence on the plumping of the pieces. Even keeping in a buffer solution at a p H of 8, at a temperature of 40" C., had little influence on the plumping. Nevertheless, bating a t 40" C. for 6 hours in a solution containing 5 grams of a commercial bate, buffered a t p H 8, restored the hide kept in alkaline solution for 1 week practically to its initial condition, as shown in the table. Hide kept a t p H 12.5 for the longer period, however, did not fall so completely. The results obtained would indicate that disintegration products of the collagen are a t least partially responsible for the incomplete falling of the hide pieces. The finding la 14 1s
Wilson, "Chemistry of Leather Manufacture," p. 109. Biochem. J . , 14, 147 (1920). Wilson, "Chemistry of Leather Manufacture." p. 111.
Vol. 19, No. 11
is fully in accord with the conclusion of MarriottI6 that removal of disintegrated cQllagen is an important function of the bating process and offers a convenient method of measuring this activity of bating preparations. It is significant that in the depression of hide plumped a t p H 11, the plumping (1.27) produced by a 4-normal solution of sodium chloride corresponds closely with the figure (1.24) obtained when hide plumped a t p H 12.5 is brought back to a p H of 8. Probably, therefore, the small influence of sodium chloride on the reduction of the plumping of hide produced by alkaline solutions is connected with the presence of partially disintegrated collagen rather than with the existence of the collagen in another form. Whereas G u ~ t a v s o n 'has ~ offered evidence of the importance to be attached to the previous history of the hide, as influencing its behavior towards tanning agents, this investigation stresses the important effect of that history on its physical properties. 18
J. Inst. SOC.Leather Trades Chem., 10, 132 (1926).
17
THIS JOURNAL, 19, 243 (1927).
Origin and Decomposition of Carbon Disulfide in Gas-Making' The Carbon-Sulfur Complex By Wilbert J. Huff and John C. Holtz DZPARTMZNT OF GAS ERGIREERING, THE JOHNS HOPKINSUSIVERSITY, BALTIXORB, MD.
T
HIS paper describes an extension of the experimental inquiry into the origin of carbon disulfide previously reported by one of the writers.2 I n the initial work it was shown that coal undergoing carbonization gave no carbon disulfide when the rate of change of temperature was very low, while coal undergoing a very rapid change in temperature gave important quantities of carbon disulfide. Under the experimental conditions of that inquiry no carbon disulfide was formed when coal gas containing hydrogen sulfide was passed over heated coke, and the carbon disulfide contained in that coal gas was partially decomposed. It was suggested that this carbon disulfide was formed in local high-sulfur regions which when heated suddenly became chemically active in the presence of a deficiency of carbon and hydrogen, thus forming carbon disulfide which was swept rapidly out of the retort by the gas stream; whereas, when heated slowly, with the aid of diffusion, the sulfur might combine to form hydrogen sulfide and non-volatile solid carbon-sulfur complexes rather than carbon disulfide. The need of additional inquiry into any explanation of the formation of carbon disulfide was stressed a t that time. Production of Carbon Disulfide during Cracking of Oil I n such an investigation the formation of carbon disulfide during the cracking of oil is of interest, because the sulfur compounds of the oil are probably uniformly dissolved in the remaining oil constituents, thus affording an experimental condition quite unlike that prevailing in coal, whose heterogeneous make-up is recognized. OIL USED-For the qualitative preliminary experiments Presented before the Ninth Annual Con1 Received July 30, 1927. vention of the American Gas Association, Chicago, Ill., October 10 to 14, 1927. 1 Huff, THIS JOURNAL, 18, 357 (1926).
which we report herewith, a gas oil having the following properties was used (unless otherwise noted) : Dala on Original Oil Color Black Viscosity 2 . 7 7 a t 77OF. ( 2 5 ' C . ) Water Trace Sulfur 3 . 6 1 7 , by wt. Specific gravity 0.9123 220' F. (104' C.) Flash point 245' F. (118' C.) Burning point Remarks: Gas Chemists' Handbook Method, 3 cc. per minute
Distillation Data PER CENTOFF By By SPECIFIC DESCRIPTION OF TEXPERATURE volume weight GRAVITY FRACTION F. c. ann -1 A 4 300-3% 149-i?? 376-400 191-204 0.4 0.3 0.769 Colorless 400-450 204-232 2.8 2.5 0.817 Colorless 450-500 232-260 9.5 8.8 0.839 Colorless 500-550 260-288 15.2 14.7 0.869 Very light yellowish green Light yellowish green 16.6 16.4 0.892 550-600 288-316 Yellowish green 1 7 . 0 0 . 9 0 4 17.0 600-650 316-343 Dark yellowish green 17.8 0.916 343-371 17,6 650-700 Reddish brown 1 9 . 1 0 . 9 2 0 371-383 18.7 700-722 1.12 Coke TOTAL
-
-
97.8
97.72
APPARATUSA S D PROCEDURE-The apparatus possessed no novel features. The cracking tube was an inclined quartz tube about 36 inches (91 cm.) long, 7/8 inches (22 mm.) internal diameter, wall thickness about 1/8 inch (3 mm.). An electric furnace heating about 12 inches (30 cm.) of tube for an interval from about 6 to about 18 inches (15 to 45 cm.) from the upper end was used. The furnace temperatures were measured by means of base-metal couples placed a t the center of the furnace between the furnace wall and the quartz tube. The upper end of the quartz tube was equipped with a suitable device for feeding the oil a t various controllable rates, and the lower end was attached to a suitable tar trap,