Variability of J
Plantation Latex SURFACE TENSION D. S. VILLARS U. S. Rubber Co., Passaic, N. J.’
[Ax AXTIQUEDROP-WEIGHTMETHODOF DETERMINING
SURFACE TENSIONOF LATEX
ITTLE has been published (4) concerning the surface tension of latex and its dependence on various plantation factors. In the course of an extensive survey of all the factors likely to contribute to latex variability, we have accumulated considerable data on this property. Although variations in the surface tension of latex are not of great commercial importance, elucidation of the effect of plantation factors on surface tension may furnish a clue to the behavior of latex in regard to other properties. At the outset of the investigation, the presumed order of diminishing importance of plantation factors affecting surface tension was:
L
Estate Factors Tapping system Part of tapping month Seasonal effects soil Age of rubber stand Tree variation (clones) Manuring treatment
Factory Factors Preservation Age of latex Temperature of latex
Experiments involving some of these factors are discussed here. The tapping systems (Table I), part of tapping month (Tables I1 and III), soil variability (Tables I1 and 111), manuring treatment (Tables IV and V), and temperature at which the latex is held (Table 11) have no important effect upon the surface tension of 1.25 per cent ammonia latex. There are differences between monthly averages (Tables I, 111, IV, and V) which may be attributed to seasonal effects or to consistent instrumental errors which change with time. The variability data are recapitulated in Table VI. These results have been analyzed statistically by the method of 1 The experimental work was carried out at the Plantation Reeearoh Department of the U. s. Rubber Plantations, Inc., in Sumatra.
368
The effect on surface tension of tapping systems, part of the alternate monthly tapping month, soil, or individual differences between trees, manuring, or temperature at which latex is aged has been found to be commercially negligible (standard deviation not greater than 0.2 dyne per cm.) for 1.25 per cent ammonia latex. Statistically significant seasonal effects are indicated but in the present experiments cannot be distinguished from consistent errors in technique which may vary with time. The surface tension of fresh a m monia latex always drops with aging. The reproducibility of the determination corresponds to a standard deviation of the order 0.5 dyne per cm.
variance analysis ( I , I,6) which is not generally familiar. This powerful tool, widely applicable in industrial work, enables one to determine the significance of the effects observed without any sacrifice in rigorousness due to the finite size of the samples.
Surface Tension Measurements The surface tension measurements were carried out on a Cenco du Nouy precision tensiometer (No. 10402). It was calibrated in accordance with the instruction bulletin by measuring the scale reading, 2, with the weights, M = 0.200, 0.300, 0.400, 0.5002, and 0.7002 gram, respectively, on the ring. The quantity y / F = Mg/2L
MARCH, 1940
INDUSTRIAL AND ENGINEERING CHEMISTRY
was computed by using 978.057 for g, the acceleration due to gravity. F was interpolated from tables of Harkins and Jordan (S) corresponding to R/r where d
=
39.1 and R 3 / V liquid
=
0.258 d / M
= density of
Calibration curves of y-z were plotted for different density liquids corresponding to water, normal latex, and cream, and all scale readings were corrected by these curves (y being the surface tension). The tensiometer was first calibrated in January, 1935, was recalibrated October 23, 1936, and again on October 24, 1937. In many cases the surface tension was determined both for the unadjusted latex and latex adjusted to 30 per cent total solids and 1.25 per cent ammonia. Many of the results reported were obtained from experiments which have already been in progress for several years a t the Plantation Research Department to test the effects of different factors on rubber yields. It was a simple matter to send bottles with preservative to the estates concerned and request samples of the latex on appropriate sampling days. In those cases where a new experiment was started, it was usually designed by the writer, with the cooperation and advice of the other members of the department.
Tapping Systems (Experiment 7) A common choice in tapping system lies between alternate monthly and alternate daily tapping. Latex obtained from the latter is very constant in quality. That from the former varies markedly throughout the tapping month; it has different characteristics when the tree is fresh from a month's rest from those when the tree is fatigued. Over a period of one year we have secured surface tension data from part of Tapping Systems Experiment KO. 7 . This experiment is located on division IVC of Serbangan estate in a white soil area of ordinary unselected seedlings planted in 1911-12. It was started on May 17, 1933; although broken up into eight replications, for the purpose of this work it may be considered as not being replicated a t all, since all latex from one treatment was bulked into one bucket, and all latex from the other treatment was bulked into a second bucket. (This means that it will be impossible to ascribe any differences discovered to treatments as distinguished from plot variability.) The areas were appropriately split so that tapping was always going on in one half while the other half was resting. Latex samples were preserved (at approximately 3-day intervals
throughout the tapping month) on the following dates: 1935: April 18, 21, 24, 27, 30, May 3, 6, 9, 12, 15; changeover to other alternate monthly tapping half, May 18, 22, 25, 28, 31, June 3, 6, 9, 12; change-over, June 18, 21, 24, 27, 30, July 3, 6, 9, 12, 15; change-over, July 18, 21, 24, 27, 30, August 3, 8, 12, 15; change-over, September 18, 21, 24, 28, October 3, 6, 11, 12; change-over, October 18, 21, 24, 27, 30, November 4, 7, 9, 12, 15; change-over, November 19, 21, 25, 27, 30, December 3, 6, 9, 12, 15; change-over, December 21, 29, 31, January 3, 6, 9, 12, 15; change-over, January 18, 21, 24, 27, 30, February 3, 6,9, 12, 15; change-over, February 18, 21,24,27, 28. (In general, the first and sixteenth of every month are rest days.) The surface tension of the latex was determined 3 days after preservation, and values averaged for each tapping month. The averages are given in Table I. TABLE I. SURFACE TENSION OF TAPPING SYSTEMS EXPERIMENT No. 7 (Serbangan IVC, white soil, 1911-12 seedlings, change-over middle of month) Surface Tension, Dynes per . m C Unadjusted-? -Adjusted"Month Clb C2C Total Av. Clb C2C Total .4v. 1935 IV-V 38.97 38.61 77.58 38.79 . , . . . . . V-VI 39.59 39.52 79.11 39.56 . . . . . . . . . . . . VI-VI1 40.44 40.46 80.90 40.45 . . . . . . . . . . . . VII-VI11 41.21 41.30 82.51 41.26 . . . . . . . . . . . . VIII-IX 40 35 40.52 80.87 40.44 40.86 41.20 82.06 41.03 4i:So 4i:97 si.'+7 4i.'89 X-XI 41.95 42.27 84.22 42.11 42.16 42.57 84.73 42.37 XI-XI1 42.26 42.70 84.96 42.48 42.93 42.96 85.89 42.95 XIILI('36) 42.46 42.68 85.14 42.57 42.49 42.79 85.28 42.64
--
.
A 1lECEIVISG
I
..
IX-x
l.Y.B.f. 3
1-11
42.17 42.29 84.46 42.23 42.47 42.40 84.87 42.44 41.73 41.92 83.65, 41.83 41.94 42.07 54.01 42.01 451.99 453.47 905.46 253.79 254.76 508.55 AV. 41.09 41.22 4i:ie 42.30 42.46 42:38 Sum of squares 37,297,5426 21,553.e3i5
11-111
Total
0
b e
T o 30% total solids and 1.2.57 ammonia. Alternate monthly (plots A1 agd A2). Alternate daily (plots C3 and D4).
On the average, this alternate daily latex appears to have s surface tension of about 0.1 dyne per cm. higher than this alternate monthly latex. A test of the results by variance analysis, however, shows them not to be significantly different; i. e., the probability is not less than 1 in 20 that the result may have just occurred by chance. The safest conclusion, therefore, is that there is no difference in surface tension demonstrated between average alternate monthly and alternate daily latex. This experiment gives a hint of the effect of season. The differences between months are highly and very highly significant,Z the latex exhibiting an uptrend in surface tension with time. The cause of these differences, however, is an open question; the following factors are confounded as a result of lack of replication-season, increasing age of rubber stand, and drift in calibration of instrument or technique of assistants running the determination. It is difficult to replicate an experiment suitably to segre-
,
BRIXGISG I S LATEX TO
369
SHED FOR PRESERVATION
%
9 The terms used are defined a s follows: Significant means t h a t t h e probability is less t h a n 1 in 20 t h a t t h e results may have occurred by chance (+), Highly significant connotes a probability of less t h a n 1 in 100 ( ++). Very highly signifcant oonnotw a probability of less than 1 in 1000(+++). Nonsignificance is designated b y a minua sien.
INDUSTRIAL AND EN(31NEERING CHEMISTRY
370
gate seasonal effects statistically (impossible to duplicate seasons reproducibly; e. g., in one year the "dry" season may be much wetter than usual and the hottest periods may occur in different months). No clear seasonal cycle is indicated here, however, since the values at the end of the year do not connect smoothly with those at the beginning. It does not seem likely that the uptrend is due to increasing age of the trees, inasmuch as a later experiment (albeit with a different type of latex) indicates no differences due to age.
Temperature Experiment Although the surface tension of latex samples averaged over a tapping month may equal that of alternate daily tapping latex, it may still vary from day to day. The preceding experiment is not suitable to test this point because of the erratic sampling procedure (unequal and different times of sampling each month). A temperature experiment proved more satisfactory and gave an indication of the variability between estate divisions (soil and individual tree variations). Samples of normal latex were preserved with 1.25 per cent ammonia from three widely separated divisions a t five different parts of a tapping month according to the following schedule :
VOL. 32, NO. 3
3. No significant differences are evident between temperatures. 4. The error on obtaining adjusted samples is greater than that on obtaining unadjusted ones (variance of former 0.06397 as compared with 0.03762 for latter), 5. adjustment (from 35-44 per cent) to 30 per cent total solids and 1.25 per cent ammonia increases the surface tension by 0.25 dyne per cm. (difference, very highly significant). This is apparently due to the higher surface tension of the water added. 6. Surface tensions of both hot and cold samples decrease with time, at the same rate. (This indicated conclusion was not tested by variance analysis,)
Latex Survey
The preceding tests showed that the variability in the surface tension of latex between estates is not demonstrable when account is taken of the error of the experiment. This was based on the results from three divisions. Another experiment was carried out involving every division on HAPM and Si Pare Pare (ninety-six in all, countingeach tap half as a separate division). Latex was sampled from each coolie's collecting bucket and preserved with 1.25 per cent ammonia. It was stirred, and two samples were bottled and sent to the laboratory. This was done daily for the first 7 days of each tap half after change-over ( t o rested fields) for 2 tapping months. The latex was tested after 3 days and 0.B. IIIA. Fjeld 127, K. P. VB, Fiqld 141, Serbangan IVA, Field 62, 3 weeks, but only the 3-day results are reported here. AB Tappinga ABC Tappingb AB Tappinga Results are averaged in Table 111. Because of the failure Oct. 17, 1935 Oct. 18 Oct. 19 22 23 24 of the estate to send in some samples (for one reason or an29 30 31 other, usually rain) and the dilution of others with rain, SOV. Nov. 6 Nov. 7 8 15 13 14 forty-nine missing values were filled in by using the general a Alternate monthly. mean. This is not a rigorous procedure but was believed at b Alternate third monthly, juat started. the time to make any significance tests wrong on the more The samples were divided after preservation; half were conservative side. (Actually this is legitimate only if the held at outdoor temperature, the other half in a 50' C. missing values occur at random. Differences tend to be overthermostat. Surface tension tests on samples, both unademphasized by this procedure and make the significance test justed and adjusted (to 30 per cent total solids and 1.25 per less conservative when the true missing value and the corcent ammonia) were made at the following ages: 1, 2, 4, 8, responding class mean are one greater and the other less than 16, 32, 46, 60, 74, 88, 102, 116, 130, 144, 158, 172, 186, 214, the general mean.) According to the variance analyses, the 242, and 270 days or until the sample was used up. The redifferences between divisions, between tapping days, and besults were averaged in various ways over the first thirteen tween tapping months are all highly significant. That they values (including 130 days age) and are given in Table 11. are not important, even though highly significant, is demonstrated in the next paragraph. The actual unweighted deviations, however, are quite TABLE11. EFFECTO F TEMPER.4TURE O X SURFACE TENSION small, and we may estimate their fiducial limits according to an ----Surface Tensic,n. Dynes/Cm.approximate method kindly described to the writer by R. A. -r -Unadjusted-Adjusted Age of hv. Warm Av. Cold Warm Cold Sa,mple, Daj Fisher. The 5 and 95 per cent value of x 2 for 95" of freedom 42.49 42.67 42.43 42.50 42.48 1 42.18 are estimated first to be 73.8 and 118 (2, page 85). Dividing 41.73 42.21 41.24 42.02 2 41.16 41.59 these values into the sum of squares of deviations between 40.22 40.95 39.49 41.42 4 39.29 40.36 39.18 38.57 39.36 39.91 38.45 R 40.15 divisions obtained in the variance analysis (42.59) gives, re38.50 37.76 38.08 38.97 38.02 38.40 16 38.39 38.04 38.10 38.35 38.42 38.16 32 spectively, 0.577 and 0.360. These are the 1 in 20 (approxi38.64 38.50 38.78 38.61 38.39 38.16 46 mate) fiducial limits to the true variance, of which the value 38.76 3 8 . 6 2 3 8 . 3 7 38.54 38.97 3 8 . 1 1 60 38.58 38.22 38.94 38.62 38.36 38.10 74 0.448 (42.59 divided by 95" of freedom) is the best estimate. 38.57 38.24 38.89 38.24 38.54 37.83 ss 38.33 38.24 37 92 37.84 38.81 37.59 102 This means that the probability is less than 1 in 20 that the 38.16 38.25 37.78 37.71 38.61 116 37.31 true variability is characterized by a weighted variance greater 38.13 37.69 38.57 38.15 37.70 130 37.24 39.20 39.21 39.21 38.95 38.96 .4v. 38.97 than 0.58, and the same probability obtains that it is characTapping Day terized by one less than 0.36, or by unweighted standard de39.43 38.95 39.32 39.53 39,06 3rd 38.84 viation (weighting, 14) greater than 0.20 dyne per cm. or 39.39 39.4s 39.16 39.13 39.30 8th 39.09 39.14 39.17 38.94 38.95 39.10 less than 0.16. Such variability is commercially negligible 15th 38.95 39.14 39.15 39.12 38.90 38.99 23rd 39.08 and shows up as significant only in the statistical analysis as a 38.69 38.96 39.14 38.77 38.80 30th 38.91 consequence of the high degree of replication. Actually the Estate Division ~. 39 05 39.07 39.02 38.72 38.82 G. B. 111.4 3 8 . 9 2 probability is still less; for the true fiducial limits lie between 39.43 39.37 39.04 39.31 39.09 K. P. VB 38.98 this approximate estimate and one using 0.448-0.266 (where 39.19 39.21 39.22 39.04 39.03 Serb. IV.4 39.02 0.266 is the error estimate) as the treatment variance estimate in place of 0.448. The latter would give 0.13 in place of 0.20 for the upper limit of the standard deviation, and 0.10 Table I1 and a variance analysis of it, as a guide to avoid for the lower limit. statistically unsupportable statements, lead to the followLikewise, the 1 in 20 fiducial limits to the deviations being conclusions : tween days may be shown to lie between a weighted (weight1. No significant differences are evident betx-een parts of the ing, 192) variance of 1.1 and 8.6, or the unweighted standard tapping month. deviation between 0.08 and 0.2 dyne per cm. The uncer2. No significant differences are evident between estates. 7
7
~
~
371
INDUSTRIAL AND ENGINEERING CHEMISTRY
MARCH, 1940
Manuring Old Seedlings OF LATEX SURVEY TABLE111. RESULTS
Estate No.
1 2 3
4 5
No. Observationsa
Av. S.,T. Filled-in,b Dynes/am.
Estate
14 13 14 13 14 13 14 13 14 13 14 14 14 14 14 14 14 14 14 13 14 13 14 14 14 14 14 14 14 13 14
40.7 40.6 40.4 40.3 40.3 40.4 40.3 40.4 40.5 40.4 40.5 40.5 40.7 40.4 40.5 40.3 40.5 40.1
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55
KO.Obser+-ationso
No.
Av. S. T. Filled-in,b Dynes/cm.
Estate N-0.
Obser- Av. S.T. vationsa Filled-in, b Dynes/cm. 14 40.5 13 40.3 14 40.8 13 40.5 14 40.5 14 40.6 14 40.5 40.3 14 14 40.4 13 40.4 14 40.5 13 40.7 14 40.7 13 40.6 14 40.8 14 40.5 14 40.6 14 40.6 40.5 14 14 40.5 14 40.6 14 40.5 14 40.6 13 40.5 14 40.7 13 40.6 14 40.7 13 40.6 11 40.4 12 40.6 11 40.4 12 40.5
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 s1 82 83 84 85
40.4 40.4 40.5 40.1 40.2 40.3 40.4 40.3 40.2 40.4 40.5 40 5 40.4 40.5 40.4 40.5 40.0 40.2 40.6 40.7 40.6 40.6 40.4 40.4 40.6 40.5 40.3 40.7 40.5 40.6 40.4 40.5
14 13 14 13 14 13 13 12 13 11 13 13 13 13 13 13 13 13 14 13 14 14 14 13
No.
The effect of manuring treatment on the surface tension of ammonia latex is ascertainable from the two following field experiments.
EXTENSION TO MANUR~KG ON REDSOILEXPERIMENT. This ex-
periment was laid out in 1911 on Goerach Batoe IB with unselected 9 seedlings. The manuring was 10 started in 1928. Although it was 11 12 replicated seven fold, it must be 13 14 considered as nonreplicated for the 15 present work since the latex was 16 17 mixed into one bulk before sam18 pling. The soil treatments are 40.3 19 20 40.3 given in Table IV. The last 21 40.3 manuring was February 24, 1935. 86 22 40.1 87 23 40.3 Tapping was on the alternate ss 56 24 40.2 89 25 40.2 57 14 monthly system, changeoverat the 90 26 40.2 58 13 firstof the month. Samplingdates, 91 40.1 59 14 27 92 28 40.5 60 14 onwhich the present averages were 93 40.3 61 14 29 94 based, were as follows: 1935, May 30 40.3 62 13 95 63 14 31 40.3 5,9, 12, 15, 20,23, 26, 29; change96 40.3 64 13 32 14 over, July 8, 11, 14, 17,20, 23, 26, Tapping No. ObserAv. S. T . Tapping No. ObserAv. S. T. 30; change-over, September 2, 5, Day vationac Av. S. T. Filled-ind Month vationse Av. S. T. Filled-in/ ---Dynes/cm.----Dynes/cm.8,11,14, 18,20,22,26,29;change1st 189 40,35 40.36 1st 652 40.21 40.22 over, November 2, 6, 8, 11, 14, 17, 2nd 189 40,25 40.26 2nd 643 40.65 40.64 21, 23, 27, 29; change-over, 1936, 3rd 190 40.37 40.37 4th 181 40.52 40.52 January 3, 5, 8, 11, 14, 17, 20, 5th 190 40 58 40.58 6th 190 40 45 40.45 23, 26, 29; change-over, March 2 , 7th 166 40.51 40.49 5, 8, 11. The monthly averages a Normal = 14. d Standard deviation = 0.037. are given in Table IV. b Standard deviation = 0.14. e Normal = 672. Normal = 192. I Standard deviation = 0.020. The variance analysis shows no significant differences due to manuring treatment. There are very highly significant differences- between months, the tainty in estimate in this case is negligible since the variance average surface tension steadily rising throughout the year ratio is high. The 1 in 20 fiducial limits to the deviations between months of tests; but whether this is due to an age of stand effect (no seasonal swings seem indicated) or due to changing technique can be estimated to lie between unweighted standard deviaon the part of the assistants carrying out the test, it is impostions (weighting 672) of 0.15 and 4.75 dynes per cm.; i. e., the chances are less than 1 in 20 that it is smaller than 0.15 sible to distinguish. NITRATEEXPERIMENT.This experiment was laid out on or greater than 4.75. Thus, the surface tension of latex may be considered conSoengei Baleh I (white soil) with unselected seedlings planted stant within a standard deviation of 0.2 dyne per cm. between in 1911. Tapping was alternate monthly with change-over a t the middle of the month. Although the experiment condivisions, and between the first 7 tapping days of the month. The estimates between the two tapping months are not close sisted of eight replications, none were effective in the present work since all latex from any one treatment was poured into enough to draw conclusions, but they are not in disagreement one bulk before sampling. The soil treatments are given in with the indications from the tapping systems experiment i Table T'. (Table I). 6 7 8
C
TABLEIv. EXTESSIOSTO
v
VI1
IX
XI 1936 I 1x1 Total AV.
B2b
39.25 40.95 40.48 42.00
39.40 41.10 40.80 42.37
42.63 42.62 247.93 41.32 Sum of squares a
b C
Unadjusted B3C
Bl"
llunth 1935
-
-
-
38.31 40.77 40.98 42.17
THE M.4NURISG ON 7
Total
_
.
_
_
116.96 38.99 122.82 40.94 122.26 40.75 126.54 42.1s
_
_
-
_
B2b
...
...
...
,..
....
... ...
4i:i2
42:78
4i:k
ii7:48
42:49
.,.
...
43.09 43.39 128.60 42.87
_
Total
B1a
_
_
. . I .
Av.
43.12 42.90 129.11 43.04 43.24 42.78 129.41 43.14 129.14 128.26 386.00 . ,. 43.03 42.75 42.89 16,5562898 _
Nonmanured (control). 4 lb. (NHa)nSO4 per tree year. 2 Ib. (Ir;Ha)rSOa 2 Ib. Diammonophos (21% X, 53.4% Ptos) per tree year
+
Adjusted B3C
7
Av.
42.90 42.63 128.16 42.72 42.71 42.35 127.68 42.56 249.28 247.21 744.42 , .. 41.36 41.55 41.20 .. 30,818.2154
_
REDS O I L EXPERIMENT
Surface Tension, Dynes per Cm.
_
TABLE v. RESULTSO F KITRATE EXPERIMENT
Month 11-111 (‘35)
,--
El0
38.89 39.60 40.65 40.50 42.19 XII-I (’36) 42.70 41.89 11-111 _ __ Total 268.42 Av. 40.92 Sum of squares
IV-v VI-VI1 VIII-IX x-XI
b 0
VOL. 32, NO. 3
INDUSTRIAL AND ENGINEERING CHEMISTRY
372
Surface Unadjusted E30 Total 39.15 117.09 39.05 38.85 118.32 39.87 40.28 40.69 121.62 40.75 40.63 121.88 42.35 42.45 126.99 42.61 42.97 128.28 -42.31 - - - - -42.18 - - - 126.38 287.22 286.92 860.56 41.03 40.99 .... 35,304.1476
E2b
---.
Tension, Dynes per Cm. .
Av. 39.03 39.44 40.54 40.63 42.33 42.76 42.13
Adjusted E3
7
El
E2
... ... ...
... ...
42164 43.04 42.06 127.64 40.’08 42.55
...
... ... ...
Total
.... .... ....
4i:61 42:jl l!%’:86 43.05 43.16 129.25 42.19 42.32 126.57 127.85 128.19 383.68 42.62 42.73 16,357.9596
....
Av.
... ... ...
42:63 43.08 42.19 42:63
Nonmanured (control). 4 Ib. (iYH&)2S01per tree year (since 1926). 5 Ib. NaNO3 per tree year (since 1919).
TABLEVI. SURFACE TENSIONVARIABILITY OF AMMONIA LATEX Degrees Standard 1 in 20 Fiducial of Free- Variance SignifiDeviation Limits dom Weighted cancea Weight Univeighted Lower Upper 1 0.0995 11 0.095 0.05 1.52 6 0.114 .06 1.82 1 0.0784 6 0.115 .07 0.27 P a r t of tapping month 4 0.0797 6 0.161 .10 .38 4 0.1552 6 2.337 192 0.110 0.08 0.2 Tapping months 10 3.0937 2 1.24 0.92 1.98 5 0.3103 2 0.394 0.26 0.82 1 59.59 672 0.298 0.15 4.75 5 6.084 3 1.42 0.95 2.97 2 0.360 3 0.35 .20 1.53 6 6.399 3 1.46 1.05 2.80 2 0,599 3 0.45 0.26 1.98 10 0.124 .07 0.55 Estate division 2 0,1534 10 0.215 .12 .95 2 0,4653 0.10 0.20 95 0.448 14 0.179 6 0.175 .lo .77 2 0.184 Manuring 3 0.148 .09 .65 2 0.066 7 0.057 .03 .25 2 0.023 43 0.093 0.05 0.41 2 0.026 15 0.019 .01 .30 Temperature 1 0.0054 15 0.038 .02 .61 1 0.0218 4 Footnote on page 369 define8 t h e signifiaanoe symbols. Factor Tapping systems
-
+++ +++ ++ +++ +++ +++ +++? -
++
Surface Tension Unadj. Ad]. Unadj. Adj. Unadj. Unadj. Adj. Unadj. Unad]. Adj. Unadj. Adj. Unadj. Adj. Unadj. Unadj. Ad]. Unadj. Adj. Unadj. Ad].
tenth the usual estimate); therefore it seems better to discount this result and charge it off as one of the times a statistical analysis is found to be wrong as a guide for drawing conclusions. (When using the 1 in 20 level of significance, one expects to draw a false conclusion 1 time in 20.) There are very highly si g nif i c a n t differences between months. We note the same uptrend that is evident in the other field experiments described.
Table
I I
Conclusions
I1 I1 I11
The preceding work was correlated and the various factors were grouped together in Table VI. None of the factors investigated exert an V I1 important effect upon the I1 surface t e n s i o n , w i t h t h e I11 IV exception of tapping months. IV Here a considerable and very V V highly significant variability I1 is indicated. As mentioned I1 before, it is unfeasible a t this time to attempt to conclude whether this is due to seasonal effects or to human and instrumental consistent errors varying over these lengths of time.
Samples were obtained on the following dates : 1935, February 22, 24, 27, March 3, 6, 12, 15 (last three dates, surface tensions not determined) ; c h a n g e-ove r , April 18, 21,24,27, May 3, 6, 9, 12, 15; changeover, June 18, 21, 27, 30, July 3, 7, 9, 12, 1.5; change-over, August 18, 22, 25, 28, September 2, 4, 6, 9, 12, 15; changeover, October 18, 21, 24, 27, 30, November 4, 7, 9, 12, 15; change-over, December 21, 30, January 3, 6, 9, 12, 15; change-over, February 18, 21, 27, March 3, 6, 9, 12, 15. The monthly averages are given in Table V. The variance analysis shows no significant differences between manuring treatments for the unadjusted samples, but a significant result from t h e adjusted latex. However, this last result was based on only nine measurements with an abnormally low estimate of error (less than one COLLECTING LATEX
I
I I11 IV IV V
Literature Cited R. A., “Design of Experiments”, 2nd
(1) Fisher,
(2)
(3) (4)
(5)
ed., London, Oliver and Boyd, 1928. Fisher, R. A., “Statistical Methods for Research Workers”, 6 t h ed., L o n d o n , Oliver a n d Boyd, 1936. Harkins and Jordan, J . Am. Chem. SOC., 52, 1751-72 (1930). Rhodes, Rz~bber Research I m t . Malaya, 1935, 119; I n d i a Rubber J., 97, 82-8 (1939). S n e d e o o r , G . W., “Calculation and Interpretation of Analysis of Variance and Covariance”, Ames, Iowa, Collegiate P r e s s , 1934; “Statistical M e t h o d s ’ , Ames, Iowa, Collegiate Press, 1938.
PRmsENTnn’before the Division of Rubber Chemistry at t h e 9 7 t h M e e t i n g of t h e American Chemical Society, Baltimore. Md
