V O L U M E 20, NO. 6, J U N E 1 9 4 8
555
(16) Faust, J., Natl. Petroleum News, 36, No. 1, R 26-9 (Jan. 5, 1944). (17) Fenske, M.R., Stevenson, C. E.. Rusk, R. 4., Lawson, N. D., Cannon, M. R., and Koch, E. F., ISD.ENG. CHEM.,ANAL. ED.,13,51-60 (1941). (18) Garner, F. H., Kelly, C. I., and Taylor, J. L., World Petroleum Cong., London Proc., 448-57 (1933). (19) Gemant, A., Trans. Faraday Soc., 32, 1628 (1936). (20) Gruse, W. A., and Livingstone, C. J., J . Inst. Petroleum, 26,41329 (1940). (21) Hersh, R. E., Lawson, N. D., Koch, E. F., Fenske, M , R., and Stevenson, C. E., Petroleum Refiner, 22, 197-205 (July 1943). (22) Hinte, J. E. van, The Delft “Baking” Test, see Bouman, C. A., J . Inst. Petroleum, 25, 774 (1939). (23) Inst. Mech. Engrs., London, I, 561 (1937). (24) Kreulen. D. J. W., and ter Horst, D. Th. J., J . Inst. Petroleum, 27, No. 213, 275 (1941). (25) Lamb, G. G., Loane, C. M., and Gaynor, J. W.,IND.ENG. CHEM.,ANAL.ED., 13, 317-21 (1941). (26) Larsen, R. G., and Armfield, F. A., Ind. Eng. Chem., 35, 586 (1943). (27) Larsen, R. G., Armfield, F. A., and Whitney, G. M., S A E Journal, 51, 310-17 (1943). (28) Larsen, R. G., Thorpe, R. E., and Armfield, F. A., Ind. Eng. ACKXOWLEDGMENT Chem., 34,183-93 (1942). (29) MacCoull, N., Ryder, E. A., and Scholp, A. C.. S A E Journal. 50, The author wishes to acknowledge the invaluable assistance of 338-45T(1942). H. Diamond and the late F. A. Armfield in the preparation of this (30) Matthijsen, H. L., J . Inst. Petroteum, 26, 72-89 (1940). paper. (31) Matthis, A. R., Rev. gkn. Blec., 21, 901 (1927). (32) hlougey, H. C., SAE World Automotive Congress, Preprints, New York (1939). LITERATURE CITED (33) Petersen, H. R., Trans. Am. SOC. Mech. Engrs., 64, 228 (1942). (1) Am. Soc. Testing Materials, Method D670-42T, A. (34) Pope, C. L., and Hall, D. A., A.S.T.M. Bull. 121 (March 1943). (2) Am. Soc. Testing Materials, Method D670-42T, B. (35) Rogers, T. H., and Shoemaker, B. H., IND.ENC.CHEM.,ANAL. (3).Anderson, B., Report International Electrotechnical CommiaED., 6,419-40 (1934). sion, Advisory Committee No. 10 on Insulating Oil, April (36) Sligh, T. H., Am. SOC.Testing MateTials Proc.. I, 461 (1927). 1935. (37) Snyder, E . A., Ibid., 23, I, 448 (1923). (4).Assaf, A. G., and Balsbaugh, J. C., Ind. Eng. Chem., 35, 909-16 (38) Stager, H., Helv. Chim. Acta, 6, 62 (1923). (1943). (39) Stark, A. R., in “Science of Petroleum” by Dunstan, A. E., (5) Baader, A., Erdol und Teer, 5 , 438, 458, 603 (1929). Nash, A. W., Brooks, B. T., and Tizard, H., Vol. 11,p. 1443, (6) Balsbaugh, J. C., Assaf, A. G., and Pendleton, W. W., Ind. Enu. London, Oxford University Press, 1938. Chem., 33, 1321-30 (1941). ;40) Talley, S.K., Larsen, R. G., and Webb, W. A., IND.ENG.CHEM., (7) Beall, A. L., S A E Journal, 40, No. 2, 48-53T (1937). = ~ N A L .ED.,17,168 (1945). (8) Bridgeman, 0. C., and Aldrich, E. W.,Ibid., 41, No. 4, 483-92 141) Thwaites, H. L., U. S. Patent 2.174.021 (Sept. 26, 1939). (1937). (42) Tichvinsky, L. M., Trans. Electrochem. SOC.,85 (preprint) (9) British Standards Institution, British Standard Specifications. (1944). DD. 279-96 in collected transactions. 148-1933. (43) Von Fuchs,’G. €I., and Diamond, H., Ind. Eng. Chrm., 34, (10) Burk, R. E., Hughes, E. C., Scovill, W.E., and Bartleson, J . D., 927-37 (1942). Petroleum Division, 102nd Meeting of AMERICAN CHEMICAL (44) Von Fuchs, G. H., Wilson, N. B., and Edlund, K. R., IND. ENQ. SOCIETY, Atlantic City, N. J., September 1941; IND.ENG. CHEM.,ANAL.ED., 13, 306 (1941). CHEY.,ANAL.ED., 17, 302 (1945). and Larson, E. C., Ibid., 15, 550-9 (1943). (45) Waters. G. W.. (11) Butkov. N., Neftuanoe Khoz., 13, 332 (1927). (46) Willey,’A. O., and Prutton, C. F., Soc. Automotive Engrs., (12) Clark F. M., and Snyder, E . A., Am. SOC.Testing Jfaterials White Sulfur Springs, W.Va., 1940. Proc., 36, 11,568 (1936). (47) Wilson, W. S.,and Kemmerer, H. R., Shell Oil Co., Inc., Wood (13) Davis, L. L., Lincoln, B. H., Byrkit, G. D., and Jones, IT’. A.. River Research Laboratory, unpublished data. Ind. Eng. Chem., 33, 339-50 (1941). RECEIVED January 25, 1945. Presented before the Division of Petroleum (14) Dornte, R. W., Ibid., 28, 26-30 (1936). (15) Evers, F.. and Schmidt, R.. m i s s . Verdflentlich. SiemensSOCIETY, Chemistry at the 108th hleeting of the A Y E R I c A N CHEMICAL Konzern. 5, 211 (1926). New York. N. Y.
have been discussed without reference to actual data correlating the laboratory tests with performance of the oils in the field. Although correlation tvould be desirable, service conditions are so varied that unequivocal conclusions cannot be drawn. The various test conditions suggested are based on an analysis of the fundamental aspects of the problem and are directed toward providing reproducibility. Although the significance of the absolute values obtained in laboratory tests may thus be questioned, they probably give a reliable relative rating of oils and an analysis of the variables that affect deterioration in practice. If i t is desired merely to knov horn well an oil behaves in an engine, probably the simplest thing to do is to run it in the engine. The way in Tvhich test results are interpreted is highly important. TT’ide experience which one laboratory may have accumulated on a particular test often accounts for the success obtained there, whereas other investigators may find the samp procedure inadequatc.
Frequency Errors in Timing with Electric Clocks R. S. CRAIG, C. B. SATTERTHWAITE,
AND
W. E. WALLACE
L’niversity of P i t t s b u r g h , P i t t s b u r g h 13, P a .
E
LECTRICALLY driven stop clocks are commonly used in
the laboratory for measuring time intervals. Obviously, the precision with which time intervals are measured can be no better than the control of the frequency of the alternating current used to drive the clocks ( 2 ) . When the commercial power line is used as a source of current, as is usually done, errors may result from the deviation of frequency from its nominal value of 60 cycles per second. Over periods of several hours or more, errors in timing, expressed on a percentage basis, usually become small, because the generators a t the power house are subject to adjustments which maintain the average frequency over a long period of time very close to 60 cycles per second. Larger errors are to be expected if the time interval is of the order of a few minutes or less.
For several years clocks operating from the power line have been used in connection with calorimetric studies a t the University of Pittsburgh. To evaluate the errors in timing resulting from this practice, the authors have performed the experiments described in this paper. EVALUATION OF FREQUENCY ERROR
Magnetically driven precision tuning forks of 60-cycle frequency are commercially available. The signal generated by the fork may be fed into a simple power amplifier, the output from which may be used to drive one or more clocks ( 1 ) . By suitable comparison of two clocks, one driven by current from a tuning fork and the other from the power line, one can evaluate the errors involved in using the latter.
ANALYTICAL CHEMISTRY
556
The measurement of time intervals using electrically driven stop clocks is subject to uncertainty if the frequency of the current supply deviates from 60 cycles per second. To evaluate the frequency error in using the commercial power available in Pittsburgh, the authors have compared two clocks, one driven by the power line, and the other by a magnetically driven precision tuning fork. The maximum deviations noted for various time intervals are less than 0.20/,. Probable errors are about 0.05%.
A Type 818 60-cycle fork, obtained from the General Radio Company, was used in this study. The average frequency of t,he fork was det,ermined by comparing a clock driven by the fork rvith the t,ime signal broadcast from the Xational Bureau of Standards. The time of observation vai,ied from 1 to 1.5 hours. The frequency of the fork Tyas determined to be 59.9860 * 0.0008 cycles per second at, 25" C. (This value differed insignificantly from the frequency given by the manufacturer.) The temperature coefficient of frequency of the fork was given by the manufacturer as approximately 0.0005~oper Centigradr degree. Tests confirmed the fact' that the influence of temperature is extremely small. The fork was mounted in a wooden box equipped with a thermometer. 4 t all times the temperature was within a few degrees (5 a t most) of 25 C., so that no correction for the t'emperature dependence of frequency seemed necessary. The clocks used in the comparisons had second hands that completed one revolution in 10 seconds. The two clocks, one driven by the power line and the other by the fork, w r e placed side by side and photographed at, 50-second intervals for a period of an hour or more. By using a photographic method any errors resulting from starting or stopping the clocks were eliminated. The exposure time was always 0.01 second. The photographs were sharp and the reading of the clocks could be estimated to 0.01 second. The experiment,s were carried out a t various times *?f the day and over a prriod of wvcral months.
DEVIATION
(SECONDS)
Figure 1. Frequency Distribution Curve for Deviations Obsened in 100 Independent (Not Overlapping) Time Intervals, Each of 10-RIinute Duration The differences between the times indicated by the trroclocks, as read from the film, were plotted as a function of time, and the resulting graphs used to provide the data from d i i c h Figures 1, 2, and 3 were constructed. The deviations occurring in 100 indcpendent 10-minute intervals mere found, and the number of periods having deviations within consecutive 0.1-second intervals were plotted against the deviations to give the frequency distribution curve shown in Figure 1. Figures 2 and 3 !\ere constructed in a similar manner from 201 periods of 8 minutes and 219 periods of 1 minute. The least counts in these cases were 0.05 and 0.01 second, respectively. The probable errors of single measurements computed (3) from the tabulated deviations are shown in Table I. Table I. Probable Errors of Single Rfeasurements Interval
Mzn. 10 5 1
KO. of Observations
100 201 219
Supplementing the information contained in Table I, .the authors have noted that the maxinium deviations observed in this study are 0.14, 0.16, and 0.l7yofor the lo-, 5-, and 1-minute intervals, respectively. Furthermore, although the data have not been quantitatively analyzed for time intervals smaller than 1 minute or larger than 10 minutes, some qualitative conclusions may be dran-n. On a percentage basis frequency errors will be about the same for time intervals of a few seconds as for time intervals of 1 to 10 minutes' duration. (For these cases the error in starting and stopping the clock would undoubtedly be the principal one.) For time intervals of approximately 2 hours or longer the percentage errors would be smaller than those listed in Table I, inasmuch as data clearly shoii that the periodic adjustment of the generators to produce an average frequency of 60 cycles per second is accomplished in less than an hour. The data shown in the frequency distribution curve have been examined n ith the idea of determining how often frequency deviations lead to errors in timing equal to or greater than 0.1%. For the number of observations indicated, 10% of the readings were in error by 0.1% or more for 10-minute intervals, 8.5% for 5minute intervals, and 13% for 1-minute intervals. It is believed that the data presented, while by no means ex-
30r
-
0 -LO
-.05
0
.05
.IO
DEVIATION (SECONDS)
Probable Error
Se.. 0 23 0 12 0 025
Figure 2. Frequency Distribution Curve for Deviations Observed in 201 Independent (Not Overlapping) Time Intervals, Each of 5-Minute Duration
% 0 038 0 040 0 042
Figure 3. Frequency Distribution Curve for Deviations Observed in 219 Independent (Not Overlapping) Time Intervals, Each of 1-Minute Duration
V O L U M E 20, NO. 6, J U N E 1 9 4 8 haustive, will give some idea of the magnitude of frequency errors expected in clocks operated from the commercial power line. It appears that for most experiments requiring reasonably precise time measurements, the clocks operated from the power line are adequate. While, strictly speaking, the results of this study are applicable only in the Pittsburgh area, the conclusions may be of interest t o others with similar problems, working under more or less comparable conditions. There is no reason for believing the control of frequency of the Pittsburgh power supply to be unique and, accordingly, the frequency errors revealed in this study are probably representative of those to be expected in using the line
557 current available in any large American city. The procedure employed can, of course, be immediately used in other areas t o ascertain the reliability of the local power source for operating electric clocks. LITERATURE CITED
(1) Fry, E. M., and Baldeschwieler, E. L., IND.ENG.CHEM.,AKAL
ED.,1 2 , 4 7 2 (1940). (2) Steiner, L. A., Chcm. Age (London), 54, 453 (1946). (3) Worthing, A . G., and Geffner, J., “Treatment of Experimental Data,” p. 167, New York, John Wiles & Sons, 1943.
RECEIVED November 26, 1947. Work supported by a basic research grant from the Office of Kava1 Research.
Spectrochemical Determination of Copper In Copper-Zinc and in Copper-Nickel AZZoys WILLIARI 31. SPICEK, Georgia School of Technology, A t l a n t a , G a . Procedures are described for the determination of copper, as the major constituent, from the ratio of the intensities of a pair of its lines. The method is suitable for the determination of copper in the range 70 to 97YG in copper-zinc alloys and 90 to 99% in copper-nickel alloys. The importance of excitation conditions and of size and shape of electrodes is shown. A possible cause of the effect is suggested.
H
ASLER and Kemp (5) have reported that copper in aluminum bronze can be determinedin the range 75 to 90% from the ratio of the intensities of two copper lines. This method of analysis has also been used by Calker (1) for the determination of bismuth and magnesium as minor constituents in various matrixes. Calker found that the same line pair can be used n-ith different matrixes, although the working curves change as the matrix element is changed. I n order to determine whether this rule also applies to major constituents, in cases other than aluminum bronze, a study was made of copper in copper-zinc and copper-nickel alloys. These systems were chosen because only solid solutions are involved and thus possible difficulties due to segregation are avoided. EQUIPMENT
The laboratory equipment included the grating spectrograph (4,5 ) , and comparator-densitometer (3) supplied by the Applied Research Laboratories. The conditions under which the spectra were obtained are 4how-n in Table I. ( 6 ) , Nultisource unit
PHOTOMETRlC PROCEDURE
The Eastman spectrum analysis S o . 1 film was processed in the ARL-Dietert film developing machine ( 7 ) which is equipped with a thermostat to hold the developing trays a t 70” I?. The films were developed, with mechanical agitation, for 3 minutes in Eastman D-19 developer, immersed in a 3Tc acetic acid solution ehort-stop for 10 seconds, and fixed in Eastman x-ray fixing bath for 30 seconds. The film was calibrated by use of a tn-o-step filter ( 2 ) using an iron arc. A gamma of 2.1 FYas obtained for the range 2850 to ‘2980 A. KO&background correction was necessary. -
trodes on a Petrey stand, with a lower electrode of the Nationa) Carbon Co. specially purified graphite. Hasler and Kemp, using Multisouice settinps of C = 5gf., L = 100gh., and R = lor., iyhich give a critically damped discharge, found that the intensity ratio of the 2882.9 h Cu I line to the 2441.6 &.Cu I liqe varied with percentage copper. Using these conditions, this variation n-as not found with the copper-zinc alloys. This indicates that the other elements present have a profound effect on the concentrational sensitivity of the lines of the major constituent. This study could not be e s t p d e d to the copper-nickei alloys, because a nirkel line a t 2441.7 A. interferes.
A study of the spectra of these samples produced by the oscillating discharge obtained n i t h the 1Iultisou:ce settings listed in Table I showed that the spark lin? 2884.383 -4.varies in intensity relative to the arc line 2882.934 A,, as the percentage of copper varies. Using the ratio of the intensities of these lines, a satisfactory working curve for the determination of copper in brass in the range 70 to 9iyowas obtained (curve A, Figure 1).
Table I.
Spectrograph settings Spark gap, 3 mm. Slit width, 60 microns Upper electrode, alloy This was the positive electrode Lower electrode, 0.25-inch giaphite rod tapered to hemispherical tip
Table 11. Composition of Copper-Zinc Allol-s Sample,
EXPERIMENTAL PROCEDURE
Conditions for Obtaining Spectra
hIultisource settings Capacitance, 2 microfarads Inductance, 50 microhenries Resistance, 0.4 ohm Output voltage, 940 volts Initiator a t , continuous Prespark, 10 seconds Exposure, 10 to 15 seconds
Copper,
NO.
%
1
65.22
The experiments of Hasler and Kemp (6) on aluminum bronze rrere repeated on a series of copper-zinc alloys, whose composition is shown in Table 11.
2
These samples were in the form of strips about 3 inches long, 1 inch m-ide, and 0.06 inch thick. These were used as upper elec-
8
3 4 5 6 7
70.61 75.37 79.96
85.22 90.01 94.2R 97.19
Zinc,
Iron,
%
7c
%
34.76
0.01 0.02 0.05 0.01 0.01 0.01 0.05 0.05
0.01 0.04 0.02 0.005 0.005 0.01 0.02 0.01
29.33 24.56 20.02 14.76 9.97 5.67 2.75
Lead,
