V O L U M E 26, NO. 4, A P R I L 1 9 5 4 made with India ink a t a spacing of 6.50 em. These strips were hung in the oven for various lengths of time, then cooled, pressed between glass slides, and remeasured. Each strip so prepared was then stretched to its original length (between bench marks); the tensile stress thus produced, which mal- be called the “recovery modulus,” was measured. As a first approximation, the recovery modulus should be equal to the set modulus, defined above, although the latter should be somewhat the smaller owing to creep during drying. Values of shrinkage and recovery modulus with a prevulcanixed natural latex (sample D of Table I\-) are plotted in Figure 14. It is significant that while the shrinkage, or loss of water, is virtually complete in 4 minutes, the recovery modulus has reached only one third of its final value a t this stage. I n other words, during the heating subsequent to driving off the water, the tension set up in a latex film on a mold would continue to increase. The present results are only preliminary. I t would appear that the tensile strength, ultimate elongation, shrinkage, and recovery modulus are the four most important factors whose interrelated change during drying and heating determines the
703 tendency of a given latex film to break on a form or mold of a particular shape. ACKNOWLEDGMENT
The authors are grateful to the Fabric Research Laboratory of Boston for permission to use the Instron testing machine, and to the General Latex and Chemical Corp. for permission to publish this work. The authors particularly express their appreciation to Dorothy Harrington and Lucy Macali Ronco, who carried out most of the dipping and tensile measurements here rrported. LITERATURE CITED
(1) Eaton, B. J., and Grentham J., J . SOC.Cheni. I n d . ( L o n d o n ) , 34, 989 (1915). (2) llaron, S. H.. Madlow, B. P., and Trinastic. J. C.. I n d . Ena. Chem., 40, 2220 (1948); 44, 1633 (1952). (3) Zwicker, B. M. G., [bid., 44, 774 (1952). RECEIVED for review May 29, 1953 Accepted January 8 , 1954 Presented before the Division of Rubber Chemistry, . h 4 E R I C 4 V C H E M I CSOCIETY, ~L Boston, Mass, Map 1953.
Colorimetric Determination of Fluoride M.L. NICHOLS and A. C. CONDO, J R . ~ Cornell University, Ithaca,
N. Y.
The bleaching effect of low concentrations of fluoride upon 18 organoferric colored complexes in acid solution was investigated to determine if any were sufficiently sensitive to be useful for the determination of fluoride. It was found that the iron(II1) complexes of three reagents, resorcylaldoxime, Sphenylsalicylic acid, and resacetophenone, gave approximately a 4% change in percentage transmittancy per 1 p.p.m. of fluoride at a pH of 2 to 3 and the proper concentration of iron(II1). However, the colored complex of resorcylaldoxime is not stable on standing for an appreciable period of time. The latter two complexes were found to be stable over long periods of time and to give reproducible and sensitive results for fluoride in concentrations from 0 to 6 p.p.m. Investigation of the effect of the presence of 13 foreign ions showed that those which form stable complexes with iron, suLh as citrate and tartrate, and also a high concentration of aluminum, interfere with the determination.
T
H E determination of fluoride ion has long been a subject of concern to analytical chemist.. Stevens ( 1 7 ) in 1936 and McKenna (11) in 1951 reviewed the history of this determination and discussed the gravimetric, volumetric, nephelometric, and colorimetric methods available. In 1936 the American Katerworks Alssociationset up a committee to study the known methods for the determination of fluoride ion in many types of waters. They selected (1, 3, 9) the Scott (16) modification of the Sanchis (16) method which uses a zirconium-alizarin reagent that is decolorized by fluoride and the color compared with standards of known concentration. The method is simple and fairly accurate, but has some disadvantages. The yellow color of the uncombined dye causes a change in hue when the zirconium-alizarin lake decreases in intensity as a result of increasing fluoride concentration, and up to 1.4 p.p.m. of fluoride the standards are unstable. Since then, other colorimetric methods ( 5 , 9, 12) have been investigated. 1
Present address, Atlantic Refining C o , Phlladelphla, Pa.
In recent years the fluoridation of community water supplies to reduce dental caries has become fairly nidespread. The fluoride ion is added by means of various compounds, generally in the range of 0.6 to 1.5 p.p.m. Therefore, a simple, rapid, colorimetric method for the determination of fluoride ion in water supplies is desirable. The bleaching effect of the fluoride ion in the colorimetric methods for determining iron(II1) in acid solution, caused by formation of the colorless fluoroferric complexes, is R ell knorT n. Thii bleaching effect of low concentrations of fluoride ion upon variou.: organoferric colored complexes in acid solution was studied to see if any M ere sensitive enough to be satisfactory for the determination of very small amounts of fluoride ion. -1survey of the literature disclosed numerous organic reagents n hich produce color reactions with iron(II1) in acid solution. Eighteen of these were selected and, a here possible, the organoferric complexes n ere produced according to the directions in the literature. The wave length of maximum absorption was determined, and then a quantitative study was made of the effect of both p H and ferric ion concentration upon the ability of the fluoride ion to cause a fading effect. The fading effect mas determined by measuring the increase in percentage transmittancy of radiant energy by the colored complex at constant p H and constant ferric ion concentration as the fluoride ion concentration R as increased. EXPERILMENT~ L
A Lumetron filter photometer bIodel402E with 2-cm. cells and filters of approximately 30 m l band width was used to measure the transmittancy. The experimental data were confined to results between 20 and 60% transmittancies to decrease the absolute errors, as recommended by Ayres ( 2 ) and Ringbom (13). A Macbeth Model A p H meter was used to adjust the acetate buffer solutions. The standard iron(II1) solution of 100 p.p.m. was prepared from reagent-grade ferrous ammonium sulfate by oxidation with bromine, and the standard fluoride solution of 2.04 p.p.m. of fluoride ion was prepared from reagent grade sodium fluoride. Buffer solutions of p H 1, 2, and 3 were prepared according to Britton ( 4 ) and were adjusted to k0.02 p H unit a t 25” C. with a p H meter. These buffers had an acetate content of about 0.2.11.
ANALYTICAL CHEMISTRY
704 The bleaching effect of the fluoride ion in the region of low concentrations (0 to 10 p.p.m.) was of primary interest. This was studied in two ways: By plotting the log percentage transmittancy against the volume of fluoride solution added to the colored organoferric complex to determine whether the change in percentage transmittancy obeyed the Lambert-Beer law. By plotting % T , - % T o us. the volume of fluoride solution added, where T , is the transmittancy with fluoride present and i", is the transmittancy with no fluoride present. The slope of this plot gives a value of A%T/AV or the change in percentage transmittancy per 1 p.p.m. fluoride. This value gives a comparison of the sensitivity of an organoferric complex to fluoride under various conditions and between various complexes. RESULTS
The first reagent studied was ferron (7-iodo-8-hydroxyquinoline-5-sulfonic acid) which Swank and llellon (18) have used for the colorimetric determination of iron. They found that with an iron(II1) concentration of 1 p.p.m. and a t a pH of 3 using a 5-em. cell, the presence of 10 p,p.m. of fluoride caused a 10% change in the color intensity. Fahey ( 8 ) and Urech (20) have invePtignted this color change as a colorimetric method for fluoride, This reagent was chosen to compare the bleaching effect with that obtained by S ~ a n kand Xellon and to see if greater sensitivity could be obtained a t any other p H or irori(II1) concentration.
Table I.
Organic Reagent
~~
!-RE SORCY L A L DOXl M E 5c
15 P F M FE 4C
e
8
30
I >
I t-
20
Filter, lllp
PH
620 513 640 ,513 490 490 .5 15 465 575 51.5
2 2 7 2 2 2 2 2 2 2 3
Fe(III),
P.P.11.
1
1 P P AI. F 0.7 0.8 1.2 2.1 1.1
2.1 2 . ,5 2 9 3.3 3.6
It was found that the greatest sensitivity was obtained a t an iron(II1) concentration of 5 p.p.m., a p H of 2, and with a 620-mp filter and 2-cm. cell, Since the percentage change depends upon the iron(II1) concentration and buffer capacity of the solution, calculations showed only t h a t a t a p H of 3 and 2 p.p.m. of iron(II1) the percentage change \vas of the same order of magnitude as obtained by Swank and Mellon. The 17 other reagents were then investigated ina similar manner. .Iminopyrine, chromotropic acid (4,5-dihydroxy-2,7-naphthalenedisulfonic acid), dinitrosoresorcinol, hematoxylin, isonitrosodimethyldihydroresorcinol, salicylaldoxime, sym-diphenylcarbazide, and tannic acid gave unstable colors or showed no bleaching effect with fluoride ion, so they nere not investigated further. For the other 10 the results sholying the optimum p H and iron(111)concentration giving the greatest changeinpercentage transmittancy per 1 p,p,m. of fluoride ion are shown in Table I. The last three reagents have the greatest sensitivity to fluoride, and these were investigated further with respect to reproducibility of results, stability, and the effect of other ions. p-Resorcylaldoxime. The complex of ferric iron and p-resorcylaldoxime (2,4-dihydrosybenzaldo~ime)w'as investigated by Chien and Shih (6). One gram of the reagent was dissolved in 25 ml. of 95% ethyl alcohol and diluted to 500 ml. with water. Two milliliters of reagent, iron(III), fluoride solutions, and buffer to total 100 ml. were used for the various iron(II1) concentrations and p H values tried. Measuring the transmittance a t 515 mp, the best results were obtained a t a pH of 3 and 15 p.p.m. of iron over the range of 0 to 10 p.p.m. of fluoride. Repetition of these experiments showed that the results varied
A
ML.
- -0RIG. BUFFER
B - -1:4 BUFFER CORRECTED C
IO
Sensitivity of Organoferric Complexes to Fluoride A%T __-
Ferron Salicylic acid Tifprrnn ...... Sulfosalicylic acid Kojic acid .4cetylacetone 6-Resorcylic acid Resacetophenone 5-Phenylsalicylic apid @-Resorcylaldoxime ~
EO -
2
- - l:4
3
BUFFER N O T CORRECTED
4
5
7
6
FLUORIDE SOLUTION
Figure 1. Effect of Buffer Concentration on Transmittancy
with the acetate content, confirming the r s u l t s of Thoms and Gantz (19) that the ferric acetate coniplea is mor? stable than the ferric fluoride complex. From Figure 1 it can be seen that the best results were obtained ith a diluted (1 to -1) buffer where the pH varied from 3.2 to 2.6 on addition of the iron(II1) solution. However, it was found that these solutions shoJyed a 2 to 3% change in transmittancy after standing in daylight for 1 hour. Because this was unsatisfactory for a colorimetric method, this reagent Tr-as not investigated further.
Table 11. Yolume S a F Solution, 111. 0
1 7
3 4 3 I
Effect of Fluoride on 5-Phenylsalicylic AcidIron(II1) Complex Pelcentage Trarisriiittancy 47.4,47.4,47.3.4i.3 5 4 . 6 , 5 4 . 4 .5 4 . 7 , 3 4 . 7 61.2,61.0.61.0,61.1 6 6 . 8 , 6 7 . 0 , f i 7 . 8 ,6 R . 8 71.6,71.8,72.?.71.3 7 6 . 7 , 73.7, 7 6 . 7 , 7 . 5 . 3 8 2 . 7 ,8 2 . 3 , 8 2 . 3 . 8 2 . 3
.iverage 47.4 94.6 61.1 67.1
(1.8 73.8 82.4
Average Deviation 10.05 1 0 .1
*O.l rO 4 10.2 10.4 JO.2
5-Phenylsalicylic Acid. The comple\: of iron(II1) and 5phenylsalicylic acid has been studied by Toe and Harvey ( 2 1 ) .
One gram of 5-phenylsalicylic acid 4-as dissolved in 100 ml. of 9570 ethyl alcohol, since the reagent precipitates in aqueous solu-
tion a t a pH of 3.3 or below. One milliliter of the reagent solution, 40 ml. of ethyl alcohol, iron(II1) solution, and fluoride solution, and buffer to total 100 ml. were used for each iron(II1) concentration and p H studied. The best conditions for fluoride sensitivity were a t pH 2 using 5 p.p.m of iron(II1) and 1 ml. of reagent; a bright blue color was obtained which was almost completely decolorized by 14 p p.m of fluoride. .4t 575 mp the percentage transmittancy was linear from 0 to 6 p.p.m. of fluoride, with a Eensitivity of 3.3 per 1 p,p,m. of fluoride, and nonlinear a t higher concentrations, The results of four runs are given in Table 11. The colors were stable over a long period of time.
705
V O L U M E 26, NO. 4, A P R I L 1 9 5 4 The effect of the piesence of various foreign ions on the color intensity of the solutions is given in Figures 2 and 3. The chief interference in water supplies would be due to aluminum and sulfate which is allowable in drinking water up to 250 p.p.m. (IO). I n addition it was found that the presence of 100 p.p.m. of calcium, 50 p.p.m. of magnesium, 5 p.p.m. of cobalt(II), or 10 p.p.m. of nickd(I1) ions does not have an appreciable effect upon the sensitivity. Resacetophenone. A 1O0G acetone solution of resacetophenone (2,4-tiihydro\yacetophenone) x as used, because the i eagent precipitated on standing from a 10% ethyl alcohol solution, as used hy Cooper (T), at a pH of 3 or below.
5
PPM
FE
The red color systems were developed by using 1 ml. of reagent, iron(II1 I , and fluoride solutions and diluting to 100 ml. with buffer solution thr same as with 5-phenylsalicylic acid. The t w t conditions for fluoiitlp scnsitivity were a t p H 2 using 10 p.p.m. of iron(TII), 1 ml. of reagent, and a 0.131 buffer.
5-PHENYLSALICYLIC
ACID
3 5 1
ML.
Figure 3.
8- 2 p C
D- 2 P P M E - 2 PPM
F- 2
I
ML.
Figure 2.
2
p
p
3
FLUORIDE
Ten parts per million of iron, 5 11.p.m. of fluoride, various amounts of aluminum as aluminum sulfate, and 1 ml. of reagent were placed in flasks and diluted to 100 ml. with the 0.1M buffer solution of p H 2. The percentage transmittancy was determined a t 10 minutes, 1 hour, and 8 hours after mixing.
p T~A R T R A T E
- 100 p P M
SULFATE
The results (Table I V ) show that the destruction of the ferric fluoride complex is a reasonably slow reaction a t higher concentrations. For the readings a t 10 minutes, there is an almost linear relationship between the change in percentage transmittancy and parts per million of aluminum. With 5 p.p.m. of aluminum there is a change of 2.5% in the '%T, - %To values.
CITRATE OXALATE
~PHO S P H A T E
4
5
6
FLUOPiDE SOLLJTION
Effect of Foreign Ions on Transmittancy
7
SOLUTION
EfTect of Foreign Ions on Transmittancj-
1-sing a 165 n i p filter, the percentage transmittancy n a s linear from 0 to 6 p.p.m. of fluoride with a sensitivity of 3.9 per 1 p.p.m. of fluoride and nonlinear up to 14 p.p.m. The results of four runs are given in Table I11 and Figure 4. The colors Fere stable in daylight for 6 hours. I n all cases, 100 ml. containing only 1 ml. of reagent and buffer was used for 100% transmittance. The effect of the presence of various foreign ions on the color intensity of the solutions is given in Figure 5 . I n addition it was found that the presence of 100 p.p.m. of sulfate, 10 p.p.m. of trtraborate, 100 p.p.m. of bromide and calcium, 50 p.p.m. of magnesium, 5 p.p.m. of copper and cobalt, and 10 p.p.m. of nickel have no adverse effect. ils in the previous case, the interference of aluminum is the most serious. -4s most water supplies are treated with alum, the effect of imaller amounts of aluminum was investigated.
Table 111. Volume N A F Solution, 111. 0 1 2 .'3
4 Table I V .
Effect of Fluoride on KesacetophenoneIron(II1) Complex Percentage Transrnittancy 3 8 . 2 , 3 8 . ? ,3 8 . 6 3 8 . 4
10 10 10 10 10 10
30.1 4z0.1
.55 , 6
h0.3
63.6 68.9 73.7 80.3
i o .1 +o.
1 rO.l
Effect of Aluniinuni on Determination of Fluoride w i t h Resacetophenone
...
10 Minutes
%T 38.7 71 . O 70.fi 70.2
... L 10 20 2A 50
38.4 47.2
4 7 . 2 , 4 7 . 2 . 4 7 . 3 .L. . 3 5 . 6 , 5 5 . 4 . 5 5 . 7 . 4.i 8 63.8,63.9,63.3,63.3 69.1,68.8,68.8,69.0 7 3 . 7 , 7 3 . 5 , 7 3 . 8 ,i3.8 8 0 . 2 . 8 0 . 3 , 8 0 . 4 ,8 0 . 3
Solution, P.P.M. Fe(II1) Al(II1) F 10 10
Average Deviation r0.2
Average
5 5 5
68.3 66.4 65.0 57.6
A%T 0.4 0.8 3.7 4.6 6.0 13.4
1 Hour
%T 37 7 70 5 70 3 67.9
66.4 63.1 60.7 44.4
A%T 0.2 2.6 4.2 7.4 9.8 26.1
8 fioui, A(;*oT 3 9 . F,
~
%T
72.0 71.8 69.9 64.8
61 .O 58.5 41.7
...
0.2 2.1 7.2
11.0 13.5 30.3
ANALYTICAL CHEMISTRY
706
45t-
Since in the previous experiments the %To was obtained with only iron present, which might account for the change in %bT, %T, values, the experiments were repeated, with readings a t 10 minutes, obtaining the %To values with both iron and the same amount of aluminum present as is present with the fluoride. The results in Table V indicate that if the %To values are obtained with aluminum present, correct %T, - %To values are obtained for fluoride.
t
40
35
A
RESACETOPHENONE PH 2 ~ O P P MFE A . -NONE I:I BUFFE
Table V.
8- - 2 p p ~PHOSPH
30
Effect of Aluminum on Determination of Fluoride with Resacetophenone
Solution, P.P.M.
Fe(II1)
Al(II1)
10
...
1:
I-
...
10 10 10
1 1 5 5
10 10
...
8
37 8 70.6
1
0 25
32.8
I
2o
c
8
4540 -
B
RESACETOPHENONE
l5 IO
0p PMTART RATE
- 20 p
p C~I T R A T E
- -20 p
p O~X A L A T E
05
E
35 -
F--lOO p
2
I
30 -
ML.
3
p
A~L U M I N U M
4
FLUORIDE
5
6
7
SOLUTION
Figure 5 . Effect of Foreign Ions on Transmittancy 25
-
sodium fluoride was added to 1 liter of the tap water to which 2 ml. of hydrochloric acid had been added to prevent the precipitation of calcium fluoride. One to 5 ml. of each of these standards was mixed with enough iron(II1) solution to give a concentration of 10 p.p.m. and was diluted to 99 ml. with the 0.1M buffer of p H 2. Then 1 ml. of a 10% acetone solution of resacetophenone was added and the percentage transmittancy determined at 465 mp.
0 k20-
'
i$! 15-
The values obtained in Table T I a ere plotted as %Tu - 907'0
>
us. parts per million of fluoride and shoaed a sensitivity of 3 . 1
I-
i$! 10-
A--ORIC.
e__
//
I: I
BUFFER BUFFER
/ 1
2
ML.
Figure 4.
3
4
5
8
over the range of 0 to 6 p.p.m. of fluoride. Two unknown solutions were measured and the fluoride concentration u as determined from this curve (Table VI). These results with reaacetophenone indicate that the percentage transmittancy values vary with different batches of reagents. If the %To and %T, values for the standards are determined with the same amount of aluminum present as in the water to be
7
FLUORIDE SOLUTION
Table VI.
Effect of Buffer Concentration on Transmittancy
Since the sensitivity to fluoride is approximately the same with both 5-phenylsalicylic acid and resacetophenone and the interference of foreign ions is less with the latter, the determination of fluoride was tried on the Cornel1 University water supply. This water contains no fluoride, but has been treated with alum. Fifty milliliters of the water was analyzed for total aluminum and iron with aluminon (14). No positive test could be obtained, and since the reagent is sensitive to mg. of aluminum, the nluniinun1 content is apparently less than 0.1 p.p.m. The determination was made by first preparing standards;
Determination of Fluoride in Water Samples with Resacetophenone
1
Standards Percentage traniiiiittance 45.0 50 2
10
56.0 58.5 61.2 63.8 70.8 74.2
Fluoride, p.p.ni. 0 2 3 4 5 6 8
Present 0 2, 6
53 5
Unknowns Percentage trans% T u - %TO mittance 45.0 ... 53.2 8.2 64.3 19.3
%Tt
- %To ...
5.2 8.5 11.0 13.5
16.2
18.8 25.8 29.2
Found
...
1.95
6.2
V O L U M E 26, NO. 4, A P R I L 1 9 5 4 analyzed and the readings taken after 10 minutes, the method nil1 determine fluoride content correctly even in the presence of up to 5 p.p.m. of aluminum. Similar modifications may be necessary in the presence of appreciable concentrations of some of the other interfering ions which are not PO common in water supplies. LITERATIJRE CITED (1) h n . Pub. Health A4ssoc., S e w York, “Standard Methods for Analysis of Water and Sewage,” 9th ed., p. 76, 1946. (2) ilyres, ANAL. CHEM.,21, 652 (1949). (3) Black, A. P.. J . Ani. Water W o r k s Assoe., 33, 1965 (1941). (4) Britton, H . T. S., “Hydrogen Ions,” p. 180, London, Chapman and Hall, 1929. ( 5 ) Bumsted and Wells. - 4 l v . k ~ .CHmf., 24, 1595 (1952). (6) Chien and Shih, J . Chincsc Chem. Soc.. 5, 154 (1937). (7) Cooper, ISD. ENG.CHEM., A N a L . ED.,9, 334 (1937). (S) Fahey, Ihid., 11, 362 (1939). (9) Ingols, Shaw, Eberhardt, and Hildebrand, ANAL. C H E l r . , 22, 799 (1950).
701 (10) Lange, X. .4.,Ed., “Handbook of Chemistry,” 6th ed., p. 757, Sandusky, Ohio, Handbook Publishers, 1946. (11) AIcKenna, Sucleonics, 8 (90.6 ) , 2 4 ; 9 (No. l ) , 4 ; No. 2, 51 (1951). (12) Price and Walker, ANAL.CHEM.,24, 1593 (1952). (13) Ringbom, 2. anal. Chein., 115, 332 (1939). (14) Rodden, C. J., ed., “Analytical Chemistry of the AIanhattan Project,” p. 388, New York, LIcGraw-Hill Book Co., 1950. (15) Sanchis, ISD. ENG.C H E Y . , k.41,. ED.,6, 134 (1934). (16) Scott, J . -4772. Water Works -4ssoc., 33, 2018 (1941). (17) Stevens, ISD. ENG.C H E M . , 2 ‘ S . 4 L . ED.,8, 248 (1936). (18) Swank and LIellon, Ihid., 9, 406 (1937). (19) Thoms and Gantz, Proc. I n d i a ~ aA c a d . Sci., 56, 130 (1946). (20) Urech, H e h . Chim. Acta, 25, 1115 (1942). ? ~ . 70, 648 (1948). (21) Toe and Harvey, J . A m . C ~ C JSoc., RECEIVEDfor rei-iew December 2 2 , 1932. Accepted January 4, 1934. Presented in part before the Division of -4nalytiral Chemistry a t the SOCIETY, Atlantic City, X. J. 122nd Meeting of the AJCERICASCHEJIICAL Froni a thesis presented by A. C. Condo, J r . , t o the Graduate School of Cornell University in partial fulfilliiient of t h e requirements for t h e degree of master of science.
Testing a Quick-Weighing Balance T. W. LASHOF and L. B. MACURDY Mass Section, National Bureau of Standards, Washington 25, D.
The increasing use of quick-w eighing balances necessitated the development of test procedures suited to such balances. Two tests have been developed, one for determining the linearity of the direct-reading scale, the other for calibrating the dial-operated weights. Estimates of the performance of the balance are obtained from each test. The computation form for the second test provides for the rapid and systematic least squares adjustment of the obsened corrections of the weights. The test procedure, H hich is presented through the use nf a numerical example and the discussion of statistical tests and estimates, provides for the calibration of the dial-operated Heights and the estimation of the performance of a quick-weighing balance,
H
OW good are modern quick-weighing balances? How much of the accuracy is sacrificed for the convenience and speed of a quick-weighing attachment? Such a device as a rider, chain, projected direct-reading scale, or set of built-in weights may contribute not only to the convenience or speed of operation of the balance but also to the variability of the balance. The test of a balance nith projected direch-eading scale (capacity 100 mp.) and a set of built-in weights (capacity 99.9 grams) for a total capacity of 100 grams is considered in this paper. The rider, chain, and other quick-neighing devices ail1 be conQidered in another paper. The tests described are designed specifically for the 100-gram Mettler Gram-atic balance, but could be adapted to other quick-weighing balances of similar design by taking into account the particular arrangement of the built-in weights. Thew special tests for the IIettler balanre were devised because the built-in weights of this balance cannot he easily removed for testing and hence must be tested along with the balance. DESCRIPTIOX O F BALANCE
Only those features of the balance related to the tests of the quick-neighing devices are described here. Built into the Gramatic balance are three sets of dial-controlled weights arranged in decades, each set controlled by a separate dial. The “tenths” dial controls four weights: one each of nominal values 0.5 and 0.2
C.
grmiJ and two of nominal valucl 0.1 grain. The weights used for cach dial setting are shown i n the following table, where the tn-o 0.l-gi~amweights are distinguished by subscripts. Dial Setting 0.1 0.2 0 3
0.4 n.5 0.6
0.7 0.8 0.9
Conihination of Weights Uscd (0.1)l ( 0 2) ( 0 2)
+ (0.1)1 + i o . 1:1 + (0.1)z io.?j + ( 0 . 1 ) ~ I0.J) + (0.2) ( 0 . 5 ) + ( 0 . 2 ) + (0.1)1 (0.5)+ ( 0 . 2 ) + ( 0 . l ) l + (0.1)z io
2)
i0.j)
Siniilarly, the “units” dial controls the four weights: 5 grams, 2 grams, 1 gram,, and 1 grams. The “tens” dial is also siniilar, 25 grams), (10 cscept that the weights come in pairs: (25 10 grams), (5 5 grams)l, and (5 5 granis)?. This is because it is necessary to balance the larger weights on the weight hanger. Thus t,he 50-gram dial setting removes two 25-gram weights symmetrically located on the hanger. Since, however, tlie two w i g h t s of a pair are alxays used together, they may be considered as one m i g h t , and therefor? the tens dial niay be treated in tlie same nianncr as the units and tenths dials. .4 fourth dial controls a 50-1iig. weight used in differential weighing which, briefly, is a n-cighing performed within the range of the optical scale-Le., without (#hangingthe dial settings. When the method of differential weighing is used, errors in the values of the dial-controlled weights do not enter into the results of a weighing. Only the variability of the balance and any rrror in the adjustment of the sensitivity will affect the results. The balance has only one pan and stirrup and two knife-edges. B counterpoise on the beam replaces the serond pan and stirrup and the third knife-edge, This feature eliminates the variability which would arise from the third knife-edge. The dial-controlled weights normally hang from the weight hanger which is attached to the stirrup. .411 neighing is by substitution-that is, n-ith the unknom-n on the pan of t,hc balance the dial-controlled n-eights are removed by the dial controls until equilibrium is achieved. The total load on the stirrup is always 99.9 grams (plus the weight of pan and weight hanger and the amount indicated by the projection scale and the 50-nig. weight). Because of the high niagnification of the optical system, the balance is particularly sensitive to tilt. The balancc should therefore be mounted on a sturdy support.
+
+
+
+