CONCLUSIONS
This system has provided an efficient method of trace analysis in the 0.1 p.p.b. range. It is now being used in routine analysis with an artific la1 smog chamber. Excellent results have been obtained in analyzing butenes rind higher-molecUlar-R-eight olefins and Paraffins, even when the total carbon concentration
of the artificial smog mixture is no greater than 3 carbon p.p.m. LITERATURE CITED
(1) Altshuuer, A. p., Bellar, T. A., Clemons, C. A., Am. I n d . Hyg.Assoc.
J. 23, 164 (1962). ( 2 ) Altshuller, A. P., Bellar, T. A., J. Air Assoc. in press. (3) Altshuller, A. P., Clemons, C. A., ANAL.CHEM.34,466 (1962).
(4) Bellar, T-1SigsbY, J. E., Jr.1 Cle~nons, C. A., Altshuller, A. P., Ibid., 34, 763 (1962). (5) West, P. W., Sen, B., Gibson, N.A., Ibid., 30, 1390 (1958). RECEIVED for review January 21, 1963. Accepted July 22, 1963. Division of Water and Waste Chemistry, 142nd Meetr ing, ACS, Stlantic City, N. J., September 1962.
Determincrtion of Available Lysine in Oilseed Meal Proteins S. RAGHAVENDAR RAO', FAlRlE LYN CARTER, and VERNON L. FRAMPTON Southern Regional Research laboratory, Agricultural Research Service,
b A simple and rc8pid method for estimation of available lysine in plant proteins is described. The protein is dinitrophenylated, hydrolyzed in acidic media, and the epsilon-DNP-lysine produced is eluted ,From an ion exchange column with a solvent composed of three parts by volume of 3N aqueous HCI and one part methyl ethyl ketone, and is determined spectrophotometrically at 435 mp.
I
is being given to plant materials especially oilseed meals, as sources of proteins of high quality for human beings and domestic animals. Because of the serious nutritional problem in other parts of the world, and our own dcmestic problem of supplying adequate proteins of suitable quality to the swine and poultry industry, interest has devel3ped in analytical means of determining the nutritive quality of proteins of .iegetable origin. The view that the nutritive value of vegetable proteins, arid of animal protein of inferior qualiLy, is directly related to the lysine in these proteins with the epsilon amino gi-oup free is well established (4, 8, 7, 16, 18). Lysine is not used unless the epsilon amino groups are free (16). Correlations between growth response of young animals and the free epsilon amino groups of lysine in the proteins they receive are invariably high, and, moreover, analyses of variance of the data indicate that a substantial portion of the variance of growth response obtained with different sources of most proteins of plant origin can tit? accounted for l ~ ythe variance of the free epsiloii amino groups of the lysine in these proteins. I n the case of commercial cottonseed meals, for example, the contribution of lysine in NCREASING ATTENT [ON
Prcsent address, Regional Research Laboratory, Hyderabad 9 A.P., India.
U. S. Departmenf o f Agriculture, New Orleans 7 9, l a .
the proteins with the epsilon amino group free to the total variance in growth of broilers and swine is so overwhelming that contributions of other seed constituents such as gossypol, carbohydrates, crude fiber, fat, etc., to the total variance in growth are completely masked (12). Several methods have been proposed for determining the free epsilon amino groups of lysine (available lysine) in foodstuffs (1, 60). The more promising methods are those involving the use of 2,4-dinitrofluorobenzene (DNFB). Lea and Hannan (15) and Schober and Prinz (20) utilized dinitrophenylation to study the reduction of epsilon amino groups during processing of milk products. Carpenter and his associates applied this procedure (5) to study the available lysine of a wide variety of animal and plant products. In the earlier methods, Carpenter et al. (8) directly measured the absorbance of the yellow, ether-extracted acid hydrolyzates of dinitrophenylated proteins in their estimation of the available lysine. I n a later modification (6), Carpenter reduced some of the error in the analyses by reacting the epsilon-dinitrophenylhydrazone-lysine with methyl chloroformate to produce an ether-soluble lysine derivative. The difference in color intensity of the hydrolyzate before reaction with methyl chloroformate and after reaction and extraction with ether was taken as a measure of available lysine. Dinitrophenol is the major by-product of the dinitrophenylation of proteins. Conkerton and Frampton (IO),by taking advantage of the difference in absorbance of dinitrophenol a t 380 nip in acid and alkaline media, were able to correct for the quantity of dinitrophenol present in the protein hydrolyzate. Other yellow substances might occur in the hytlrolyzates of some dinitrophenylated proteins (13) and introduce errors in a colorimetric estimation of
available lysine. I n addition, the brown humin pigments that invariably occur in acid hydrolyzates of proteins may also be expected to contribute to the error in some of the earlier procedures. I n a more recent study of dinitrophenylated cottonseed meal proteins, Frampton (12) found differences in the lysine content of the oilseed proteins before and after dinitrophenylation. The lysine lost on dinitrophenylation was recovered as epsilon-DNP-lysine. Di-DYP-lysine was not detected in any of the hydrolyzates. Presumably the free epsilon amino groups of lysine in the oilseed proteins will react with DXFB, while those that are bound chemically will not. This difference in the lysine content, as determined by the ion exchange technique of Spackman, Stein, and Moore (22) is taken in this paper as the reference in establishing the validity of the procedure reported here for available lysine. Many of the sources of error which are found in the earlier methods are eliminated in the procedure described herein where epsilon-DXP-lysine is separated from dinitrophenol and other yellow derivatives of the reaction between DNFB and the meal proteins, as well as from the brown humin products of acid hydrolysis, through the use of an ion exchange column that is developed with a mixture of methyl ethyl ketone and aqueous HCl(91). EXPERIMENTAL
Apparatus and Reagents. All absorbance determinations were made at 380 or 45.5 mp (unleis otherwise stated) with 3 Beckriian spectrophotometer, Model B. Ten-millimeter cuvettes were 1ised . The chromatographic column was constructed from IO-mm. i.d. glass iribiiig 1.5 to 16 rrn. i i i lcfiiigtli, \ k i t h ~1 medium-porosity sintercd glass disk and a Teflon stopcock fused in the VOL. 35, NO. 12, NOVEMBER 1963
1927
bottom end. A 125-ml. bulb, with an 18/7 ball joint, (ball joint on top of bulb) was fused to the top end of the tube to serve as a reservoir. The ball and socket joint served to facilitate the application of air pressure to the top of the column during the course of the chromatographic separation. The reagents used were: 10% aqueous sodium bicarbonate; anhydrous ethanol; peroxide-free diethyl ether; 2,4-dinitrofluorobenzene; 3N aqueous HCl; constant-boiling aqueous HCl; redistilled methyl ethyl ketone, free of yellow color. The eluting solvent, composed of 3 volumes of 3N aqueous HC1 and 1 volume of methyl ethyl ketone, was prepared fresh for each elution. The absorbance was almost identical with that of 3N HCl a t 360 and 435 mp. In the preparation of the column, a wet slurry of the resin (IQ), size B Arnberlite IR-120 in the sodium form, was poured into the top of the tube until a column of about 6 cm. was formed. This resin was allowed to settle about 1 cm. under gravity and then packed The under a slight air pressure. column was then washed with 3N HCl until the effluent was acidic and the final height was adjusted to about 5.5 rm. The column was kept moist a t all times. After each determination the column wa? regenerated by washing it with water and with 1N aqueous SaOH until the effluent was alkaline. I t was then washed with water and finally with 3147 aqueous HC1. Procedure. The defatted meal sample was ground to pass through a 40-mesh screen and a 0.5-gram 5-mg. sample, containing 30 to 50 mg. of N, w2s placed in a 1-liter roundbottomed flask together with a few glass beads. Care was taken to deposit the sample a t the bottom of the flask. Ten milliliters of bicarbonate solution were added, care being taken that no meal adhered to the side of the flask. The contents were then thoroughly mived by gcntle swirling, and the suspension was permitted to stand for 10 minutes. h solution of 0.3 ml. of DXFB in 10 ml. of ethanol mas then added and the contents of the flask were thoroughly mixed by gentle smirling. The side of the flask was rinsed with 3 ml. of absolute ethanol and then the contents were shaken in subdued light for two hours on a shaker with a wrist-like motion. Previous work (10) has established that 2 hours reaction time is sufficient for substantially complete reaction between DXFB arid the free epsilon amino groups of lysine. The dcohol and most of the water were removed by evaporation under an air stream, and the residue u as evtracted with four 50-ml. portions of anhydroua, peroxide-free diethyl ether. The ether in each case was removed from the residue by decantation and the residue in the flask was dried a t ambient temperature by aeration. Two hundred milliliters of constant-boiling aqueous HCl nere added to the flask, and the resulting mixture was boiled overnight a t the refluu temperature, cooled and then
*
1928
e
ANALYTICAL CHEMISTRY
filtered through a sintered glass funnel directly into a 250-ml. volumetric flask. The filtrate, and washings from the filtrate, were made up to 250 mi. with distilled water. An aliquot (2.0 to 4.0 ml.) of the hydrolyzate containing from 0.1 to 0.3 mg. of epsilon-dinitrophenyllysine was added to the column, care being taken not to disturb the surface of the resin. The hydrolyzate was driven into the column under air pressure until the meniscus coincided with the surface of the column packing. The sides of the reservoir were washed down with three 1-ml. portions of 3N HC1, each being driven into the column in the manner described for the sample. The column was then developed with 3N HC1, the total volume, including the HCl used to wash down the reservoir, being 40 ml. The air pressure was adjusted to induce a flow of 12 to 15 drops per minute. Dinitrophenol, and other yellow components, came through the column during the elution with HC1. The eluting solvent was then changed to methyl ethyl ketone-3N aqueous HCl mixture when the meniscus of the 3N aqueous HC1 (initial eluting solvent) reached the surface of the packing. Forty-five milliliters of this second eluting solvent were added, and again the flow through the column was assisted by means of air pressure as described above. The progress of epsilon-DNP-lysine down the column, as well as the methyl ethyl ketone-3N aqueous HCl solvent front, can be followed visibly. With the 5.5-cm. column, the first 12 to 13 ml. of the second solvent coming through the column were discarded. At this time the epsilon-DKP-lysine band usually had reached a point about 0.5 cm. from the bottom of the column and the collection of the effluent was initiated. The effluent was filtered through a Whatman No. I or No. 2 filter paper directly into a 25ml. volumetric flask. About 22 ml. were collccted, and the filter and funnel were then washed with enough of the solvent to bring the volume up to 25 ml. The absorbance of the contents of the flask was determined at 435 mp, using the solvent mixture as the reference solvent. Available lysine, expressed as grams of lysine per 16 grams of nitrogen, was calculated from these data. An absorptivity, a, of 14.8 a t 435 mp (10-mm. light path and concentration of 0.1 gram/per cent) for epsilonDNP-lysine hydrochlorideSHz0 was calculated by the method of least squares (15 degrees of freedom). RESULTS A N D DISCUSSION
The recovery of epsilon-DNP-lysine added to the column is indicated by the data reported in Table I, and is adequate for a spectrophotometric determination of available lysine. We observed no loss in epsilon-DNP-lysine when it was boiled in 6N aqueous €IC1 for as long as 96 hours. Also the value for epsilonDNP-lysine in cottonseed meal ww observed to be independent of time of
Table 1.
Recovery of Epsilon-DNPLysine from Columns
Added,
Recovered,
mg. 0.104 0.138 0.200 0.345 0.400
mg. 0.105 0.139 0.200 0.341 0.395
Recovery,
%
101 101
100 08.8 98.8
hydrolysis from 6 to 36 hours. Moreover, epsilon-DNP-lysine added to dinitrophenylated cottonseed immediately before hydrolysis was completely recovered. Good resolution on the column of epsilon-DNP-lysine in acid hydrolysates of dinitrophenylated protein is obtained, as may be seen from the data plotted in Figure 1, where the volume of eluting liquid is plotted against its absorbance a t 360 mp. Dinitrophenol was identified as a component in the first peak, the components making up the second peak remain unidentified, while the component in the third peak was identified as epsilon-DNP-lysine. A known sample of dinitroaniline was eluted a t the same position as the second component and therefore, may be present in that fraction. The fraction containing the third component was evaporated to dryness, and an aliquot taken up in pH 2.2 citrate buffer and subjected to analysis on the 15cm. column according to the procedure of Spackman, Stein, and Moore (22). The major constituents detected in the eluate were ammonia, epsilon-DNP-lysine, and O-DNP-tyrosine (13). The absorption of 0-DNP-tyrosine a t 360 mp, although very weak, would contribute an error were the determinations carried out a t that wavelength. I t does not absorb a t 435 mp, however, and the spectrophotometric deterrninations should be made at the longer wavelength. The methyl ethyl ketone-3N HCl eluates comprising the epsilon-DNPlysine and 0-DNP-tyrosine fractions from several determinations were combined and evaporated to a small volume. An aliquot was subjected to paper chromatography using the tertiary amyl alcohol-phthalate buffer system of Blackburn and Lowther (3) for development. A very strongly predominant yellow spot due to epsilon-DKP-lysine appeared on the chromatogram. A very faint yellow spot, due to an unidentified component which migrated ahead of epsilon-DNP-lysine, was the only colored component, in addition to the lysine derivative, that could be detected. Studies on acid hydrolysis of proteins have been carried out by various in-
Table II. Reproducibility of Available Lysine Determinations with Cottonseed Meal K-9191" Sample Replicate values for available No. lvsine. eramsll6 erains S 4.03 4.11 4.00 4.04 3.93 4.02 3.97 4.01
3.37 3.94 3.91 3.97 3.90 3.88 3.87 3.87
4.03 4.11 4.01 4.04 3.85 3.87 3.90 3.90
0 Prepared using the scetone-hexanewater mixture of King, Kuck, and Frampton (14).
ELUATE VOLUME IN ml.
Figure 1.
probably also of other oilseeds. When heat is avoided in processing of cottonseed, all of the meal lysine is available lysine (16). The analyses of variance of the regression of available lysine data in Table I11 obtained by the elution method on that obtained by the "difference" method show that the ratio of the mean square due to the regression to the residual mean square is about 55. For eight and one degrees of freedom, the odds for correspondence between the data are more than 1000:1. The method lends itself to routine analysis since in this laboratory it has been possible for one operator to carry out simultaneously eight independent determinations.
Separation of epsilon-DNP-lysine
vcstigators in efforts to minimize losses in amino acids that occiir. In the use of HCl as the hydrolyairg agent, a large ratio of acid to protein should be employed (11). While the relationship between the ratio of acid to protein and the loss of the lysine derivative was not studied in this investigation, losses of epsilon-DXP-lysine wwe greater when the hydrolysis was c a ~ i e dout with a small volume of 8.111~HC1 (5) than when the hydrolysis was carried out with 200 ml. of constant-boiling HCl per 0.5 gram of meal. h i addition, more humin is produced when the ratio of acid to protein is low. hcc~ordingly,a ratio of 200 ml. of constant-boiling IlCl per 0.5 gram of meal was used for the hydrolysis in obtaining the data reported here. The reproducibility of the results is indicated by the data iecorded in Table 11. A cottonseed meal that waa prepared by extracting raw cottonseed flakes with a mixture ol acetone, hexane, and water, in accordan:e with the directions of King, Kuck, arid Frampton (14) was used. Eight succwive 0.500-gram samples of the ground meal (40-mesh) were weighed into round-bottomed flasks, subjected to di zitrophenylation, and finally to hydrolysis, as indicated above. The epsilon-DNP-lysine in each hydrolyzate was then determined in triplicate. The over all s t a n d a d deviation from the mean calculated was ~k0.038. Analyses of variance for these data show that the mean square of the variance for between samples is (3.003, while that for within samples is 0.00078. The ratio for degrees of freedom of 7 and 16, indicates that the oddt; are of the order of 100: 1, that the cornlined errors introduced by weighing cut the separate samples, carrying out the dinitrophenylation and hydrolysis are greater than the errors in determining the epsilonDNP-lysine present in tlle hydrolysates. The values obtained with the procedure described in this paper are com-
pared in Table 111 with those obtained on ion exchange chromatography (22) before and after dinitrophenylation of a series of cottonseed meals that were previously subjected to exhaustive nutritional study (12, 16). Included also are data for one sample each of sesame, peanut, and soybean meals. The grand average for the available lysine determined by ion exchange chromatography is 3.04; that obtained by the elution technique described in this paper is 2.95. The coefficient of correlation calculated for these two sets of data is 0.990. To make the picture more complete, we are including in Table I11 data for the total lysine contents of the several meals. In explanation of the total lysine data, it may be noted that while the total lysine content of cottonseed proteins is essentially constant a t 4.2 grams per 16 grams of nitrogen, the total lysine contents of cottonseed meals is dependent upon the thermal conditions employed in processing the seed for oil. A substantial part of the total lysine may be lost (17). Losses in lysine are also attendant upon the processing of sesame seed (9) and of peanuts (2) and
Table 111.
Meal
ACKNOWLEDGMENT
The authors gratefully acknowledge the assistance of Austin C. F. Maqon in preparing the columns, and valuable suggestions offered by Wilda IT. Martinez during the course of the study. LITERATURE CITED
(l)qBalign,T3. P., Bayliss, AI. lC., Tqrnan, C. h l . , Arch. Baociiern. Ijiophils. 84, I (1959). (2) Bensabat, L. S., Franipton, V. L.,
Lysine in Oilseed Meals
Avai1al)le lysine, methyl ethyl Lysine from ion exchan e column, ketone-HC1 grame/l6 grams f f ( 1 9 ) elution method, Before After grams/l6 grams dinitrodinitrophenylation phenylation Difference N
Cot't'onseedNo. 106 3.74 Cottonseed KO. 108 0.81 CottonReed KO. 109 4.01 Cottonseed No. 110 3.04 Cottonseed No. 111 4.04 Cottonseed N o . 112 4.23 Cottonseed No. 113 2.93 2.39 Sesame Peanut 3.33 Soybean 6.15 Grand Average Coefficient of correlation = 0.990
0.42 1.08 0.92 1.14
1 .oo
0.84 1.54 0.21 0.11 0.21 ~
3.32 2.33 3.09 2.50 3.04 3.39 1.34 2.18 3.22 6.94 -
3,2!) 2.22 3.06 2.45 3.06 3.40 1.55 2.12 3.23 5.42 ~
3.04
VOL. 35, NO. 12, NOVEMBER 1963
2.98
1929
Allen, L. E., J . Agr. Food Chem. 6 , 778 (1958). (3) Blackburn, S., Lowther, A. G., Biochem. J . 48, 126 (1951). (4) Boyne, A, W., Carpenter, K. J., Woodham A. A., J . Sci. Food Agr. 12 832 (ld61). (5) bar enter, K. J., Biochem. J . 77, 604 (1960p. ( 6 ) Carpenter, K. J., Proc. Nufr. SOC. Engl. Scot. 17,91 (1958).
( 7 ) Carpenter, K. J., Ellinger, G. &I., Poultry Sci. 34, 1451 (1955). (8) Carpenter, IC. J., Ellin er, G. M., Munro, M. I., Rolfe, E. J., brit. J . Nutr. 11 162 (1957). ( 9 ) barter, F. L., Cirino, V. O., Allen, L. E., J . Am. Oil Chemisfs' SOC.38. 148
(mi).
(10) Conkerton, E. J., Frampton, V. L., Arch. Biochem. Biophys. 81, 130 (1959).
(11) Dustin, J. P., Czajkonska,
C., Moore, S., Bigwood, E. J., Anal. Chim.
Acta 9, 256 (1953). (12) Frampton, V. L., U. S. Dept. of
Agriculture, New Orleans, La., unDublished data. 1963. ( I $ ) Handwerck,' V., Bujard, E., Mauron, J., Biochem. J. 76, 54P (1960). (14) King, W. H., Kuck, J. C., Frampton, V. L., J . Am. Oil Chemists' SOC.38, 19
(1961). (15) Lea, C. H., Hannan, R. S., Biochim. Biophys. Acta 4, 518 (1950). (16) hlann, G. E., Carter, F. L., Frampton, V. L., Watts, A. B., Johnson, C., J . Am. Oil Chemists' SOC.39, 86 (1962). (17) hlartinez, W.H., Frampton, V. L., J . Agr. Food Chem. 6,312 (1958). (18) Martinez, W. H., Frampton, V. L., Cabell, C. A., Ibid., 9, 64 (1961). (19) Moore, S., Spackman, D. H., Stein,
W.H., ANAL.CIIEM.30, 1185 ( 1 9 5 8 ) ~ ~ (20) Schober, R., Prinz, I., Milchwisse schuft 11 , 466 (1956). (21) Seki, T., J . Biochem. Tokyo 47, 253 (1960). (22) Spackman, D. H., Stein, TT'. H., Moore, S., AKAL.CHEW30, 1190 (1958). RECEIVED for review October 29, 1962. Accepted August 5, 1963, Division of Analytical Chemistry, 142nd Meeting, ACS, Atlantic City, N. J., September 1962. S. Raghavendar Rao was granted a research associateship by the United Nations Children's Fund to participate in this research. The use of a company name and/or product by the U. S. Department of Agriculture does not imply approval or recommendation of the product to the exclusion of others which may also be available.
Determination of Hydrocarbons in Crude Cap itIa ry -Column Gas Chrorna tog ra phy
Oil by
RONALD 1. MARTIN and JOHN C. WINTERS Research and Development Deparfment, American Oil Co., Whiting, Ind.
b A capillary-column method has been developed for determining saturates through C7 and alkylbenzenes through Clo in crude oils. The hydrocarbons to be determined are separated from the crude with a packed prefractionator column, collected in a liquid-nitrogen trap, and then released into either of two capillary columns through a stream splitter. Components through C7 are well resolved in 4 hours on a 500-foot capillary column coated with 1 -octodecene. Alkylbenzenes through Clo are resolved on an 8OO-foot column coated with polyethylene glycol. Uncertainties in the results generally are less than 6% relative.
D
of composition is needed to characterize crude oils fully and t o provide insight into the formation of petroleum. Conventionally, individual components are determined by analyzing narrow-boiling distillation fractions with a variety of techniques (1-8, 6, 8, 14-16, 18, 20). dlthough results are usually satisfactory, these analyses are prohibitively lengthy and have been made on only a few crudes. Thus, generalizations on crudeoil composition had to be based on limited data. Packed-column gas chromatography has recently been used for determining individual components in crude oil (7,9,IP, 17,19). This approach, which has many advantages over distillation, has given analyses as far as CT (9), but resolution is less than ideal even with runs on three different columns. A new method was desired that would ETAILED KNOWLEDGE
1930
ANALYTICAL CHEMISTRY
provide a n improved resolution of components through C7 and would be faster. There also was a need to extend the analyses to alkylbenzenes through CIo, because these components have been determined in only two crudes-those both by distillation (14, 28). Capillary columns, which have excellent resolving powers (4, 6),were investigated. This led to the development of a new method for determining saturates through C, and alkylbenzenes through Clo. The method is accurate, can detect trace components, and does not need a prior distillation. The relative speed of the new method has allowed many crude oils to be analyzed. Analyses with this method of 24 crude oils (IO,11) have led to new generalizations on composition, and have provided a basis for speculation on petroleum origin. EXPERIMENTAL
The apparatus is diagrammed in Figure 1. Hydrocarbons of a selected boiling range are separated from the crude oil by the prefractionator column and are collected in the liquid-nitrogen trap. The prefractionator, a short packed gas chromatographic column, separates approsimately by boiling point. After the desired boiling-range fraction has been trapped, the prefractionator is bypassed by changing the position of the 4-way valves. The trap is then warmed and the hydrocarbons are carried into the capillary rolumn through the stream splitter. The individual components are detected by hydrogen-flame ionization as they emerge from the capillary column. Volatile compounds remaining on the
orefractionator are removed bv backflushing. For hydrocarbons through C7, a caDillarv column coated with 1-octad&ene " is used. For alkylbenzenes through Clo, one coated with polyethylene glycol is used. Hydrocarbons through C7. Numerous liquid phases were tested for separation of hydrocarbons through C7. As shown in Figure 2 for a typical crude oil, excellent separations were obtained with a 500-foot by 0.010-inch stainless-steel capillary coated with 1-octadecene. Abbreviations on the chromatogram are identified in Table I. Except for cycloheptane, which has been reported as a trace component in only one crude (id), all hydrocarbons through C, and the 10 lowest-boiling Cs's are well resolved. Methane and ethane, however, are not determined because of nonquantitative collection in the liquid-nitrogen trap. The column was operated a t 30" C. with a helium exit flow of 0.85 cc. per minute at a gauge pressure of 35 p s i . Just as on silicone-coated capillary columns ( I S ) , elution positions of some hydrocarbons change with temperature on the octadecene column. A temperature of 30' C. gives optimum resolution. An increase or decrease of only 3' C. causes a t least one pair of compounds to elute together. With a n increase in temperature, cycloparaffins and aromatics are retained longer relative to paraffins. Alkylbenzenes through C,O. Dehermination of alkvlbenzenes through Clo is complicated b y interfering sat& rates that accompany the alkylbenzenes from the prefractionator column. These saturates would hopelessly cover the alkylbenzenes if a column separating in boiling-point order were used. However, this interference is minimized by