May, 19501 SUTTON AND MARKLAND : HORTVET THERMOMETERS 251 The Standardisation of Hortvet Thermometers BY R. W. SUTTON AND J. MARKLAND (Read at the meeting of the North of England Section on February 26th, 1949) SYNoPsIs-Further evidence is brought to suggest that the use of sucrose solutions forms a satisfactory basis for the standardisation of Hortvet thermo- meters. Results obtained by the use of corrections from certificates supplied by the National Physical Laboratory for two thermometers are recorded. These are less concordant than those obtained by using corrections following standardisation of the thennometers with sucrose solutions. When the National Physical Laboratory corrections for these thermo- meters were used, the freezing-point depressions for sucrose solutions were invariably smaller than those recorded by Hortvet for 7 and 10 per cent.w/v solutions, and those calculated for intermediate strength solutions by Stubbs and Elsdon. As a result, freezing-point depressions of milk measured with these thermometers by the method proposed by Aschaffenburg and Hall are at least 0.005" C. smaller than would be recorded by the method used at present. IN a previous paperl we have discussed possible sources of error in the determination of the freezing-point of milk by the Hortvet process, emphasised the need for careful work with sucrose solutions, where there is somewhat greater difficulty in obtaining concordant results, and recommended the adoption of a standard technique in making the tests. Aschaff enburg and Hall,2 working with a platinum resistance thermometer, obtained erratic results with sucrose solutions, and record that with a 10 per cent.w/v sucrose solution there was a spread in the results amounting to 0.017" C. They concluded that standardisa- tion of Hortvet thermometers by the use of sucrose solutions was unsuitable and suggested that the thennometers should be standardised at a recognised testing institution to an accuracy of at least k0.002" C. on the International Temperature Scale. Such a system would have obvious advantages and equally obvious disadvantages but, whatever the balance in these, it is our opinion that the proposal ought not to be adopted until there is sufficient evidence that it would lead to more consistent results and that it would not necessitate a change in the limits hitherto adopted as typical of genuine milk.In our earlier paperl there is evidence that differences in replicate detenninations with sucrose solutions need not be large, and in this paper we bring further evidence to suggest that, by using a carefully standardised technique, large divergencies need not arise. We also give our experience in using thermometers for which certificates were obtained from the National Physical Laboratory. The results include determinations with three thermometers, Nos. 25918, 31320 and 425505. Certificates of examination have been supplied by the National Physical Laboratory for the latter two. Thermometer No. 25918 was in regular use in our department between 1933 and 1948 but, since no certificate for this thermometer was obtained, results with it are included only in Table I which deals with the reproducibility252 SUTTON AND MARKLAND: THE STANDARDISATION OF [Vol.75 of determinations on sucrose solutions. Results with thermometer No. 425505 are not included in Table I since only one comprehensive standardisation of this thermometer has yet been made. RESULTS WITY SUCROSE SOLUTIONS FOR STANDARDISATION Following the paper by Aschaffenburg and Hall2 we have made a survey of our results with sucrose solutions in the standardisation of two thermometers during the last five years. These results are given in Table I. TABLE I RESULTS IN STANDARDISATION OF THERMOMETERS WITH SUCROSE SOLUTIONS Thermometer 259 1944 . . 1944 .. 1947 .. 1948 .. 1948 . . Totals . . 7 per cent.w/v & No. of Average deter- correc- minations tion] O c. 8 . 18 0.0059 . 42 0.0071 . 12 0.0076 . 12 0.0076 . 84 0-0070 Thermometer 31320 1944 .. . . 12 0.0006 1947 .. . . 8 0.0032 1948 .. , . 12 0.0026 1948 .. . . 10 -0*0002 Totals . . * .. 43 0.0015 9 per cent. w/v 1 No. of deter- minations Thermometer 25918 1944 .. .. 10 1944 ,. .. 1947 .. .. 1948 .. .. 18 1948 .. .. 12 Totals - . . . 40 Thermometer 31320 1944 .. .. 16 1947 . . .. 1948 .. .. 12 1948 .. .. 10 Totals . . .. 38 1 Average correc- tion] O c. 0-0091 Q 0.0114 0.0110 0.0107 0.0041 0,0027 0-0015 0-0030 7.5 per cent:. w/v 8 per cent. w/v 8.5 per cent. w/v +--7 - & No. of Average No. of Average No. of Average deter- correc- deter- correc- deter- correc- minations tion, minations tion? minations tion, O c.O c. O c. 12 0.0094 13 0.0101 12 0.0099 12 0.0080 12 0.0102 12 0.0120 12 0-0086 12 0.0099 12 0.0105 36 0,0087 37 0.0101 36 0.0108 12 0.0025 12 0-0028 15 0.0022 12 0.0027 12 0.0023 12 0.0034 10 0-0018 10 0,0021 9 0*0016 40 0.0026 34 0.0024 36 0.0024 6 0.0028 9 5 per cent. w/v 10 per cent. w/’v +--7 - No. of Average No. of Average deter- correc- deter- correc- minations tion, minations tion, O c. O c. 12 0.0104 12 0*0110 7 0.0101 12 0.0117 12 0.0123 24 0.0113 31 0.0113 17 0-0041 14 0-0038 12 0.0026 10 0.0032 10 0.0023 10 0.0016 39 0-0032 34 0.0030 The average correction is that to be added to the depression determined in order to give the depressions adopted by Hortvet for 7 and 10 per cent. w/v solutions. and those calculated by Stubbs and Elsdons for intermediate strength solutions. It will be seen that in our work we do not encounter the large irregularities reported by Aschaffenburg and Hall.Usually a repeat standardisation with any sugar solution (involving about 10 or 12 determinations) has led to very little alteration in the correction previously used. The table does not include details of the spread of individual results, but we can state that with 10 per cent. sucrose solution the extreme figures in the 31 deter- minations for thermometer No. 25918 differed by 0.006” C. and those for the 34 determinations with thermometer No. 31320 differed by 0403°C. These results were for the most part obtained using sucrose of AnalaR quality, but a few of them were obtained with sucrose of “pharmaceutical quality” when the AnalaR reagent was not obtainable. Careful analysis of this sucrose failed to reveal the presence of any impurity, and polarimetric examination indicated that it was indistinguishable from 100 per cent.pure sucrose.May, 19501 HORTVET THERMOMETERS 253 We have encountered one sample of sucrose of "laboratory reagent" grade which appeared to be unsuitable. This particular sample was first used in the preparation of a 9 per cent. w/v solution during a complete re-standardisation of thermometer No. 31320. Ten measure- ments of the freezing-point depression were made. These were in excellent agreement in themselves, but they indicated a depression smaller by 0.003" C. than was expected from the previous standardisation. Up to this point the examination of solutions of other strengths, prepared from the last of the AnalaR quality sucrose, had indicated little need for change in the corrections.The sucrose was therefore suspected. It was found to be free from moisture and ash, but on polarisation we recorded an ccB of 67.3. The results were therefore consistent with the presence of a sugar of higher molecular weight and higher specific rotation than sucrose and, although we have not as yet obtained further evidence, we think that this particular specimen may have contained a small proportion of raffinose, which is known to be a constituent of sugar beet. Soon afterwards further AnalaR sucrose was available and the freezing-point depression of a 9 per cent. w/v solution was found to be in excellent agreement with previous work.Further, we have on many occasions examined samples of milk in two different cryoscopes fitted with thermometers which had been standardised with sucrose solutions. The agreement has been good. Our records for the last 52 samples so examined show that in 45 samples the results did not differ by more than 0.002" C., in 7 samples there was a difference of 0.003" C., and on no occasion was a greater difference found. Whilst, therefore, we have ourselves been concerned to point out the possibility of errors in working with sucrose solutions, we consider that these can be kept quite small and that the method forms a satisfactory basis for the standardisation of Hortvet thermometers. RESULTS USING CORRECTIONS FROM N.P.L. CERTIFICATES After receiving certificates from the National Physical Laboratory for thermometers Nos.31320 and 425505, we applied the appropriate corrections to our readings obtained in the examination of milks and it was soon apparent that the resulting depressions were distinctly smaller than those which were recorded by application of corrections from standardisation of the thermometers with sucrose solutions. With thermometer No. 31320, the depressions were smaller by 0.005" to 0.007" C., and with thermometer No. 425505, usually by 0-005" to 0.009" C., but occasionally by as much as 0.013" C. when measuring the smaller depressions of watered milks. The explanation for these differences was readily found when corrections from the N.P.L. certificates were used in determinations of the freezing-point depressions of sucrose solutions.(1) Sucrose solution, % w/v 7.0 7.5 8.0 8.5 9.0 9.5 10.0 TABLE I1 RESULTS WITH SUGAR SOLUTIONS Readings corrected from N.P.L. certificates (2) (3) (4) (5) (6) Thermometer 31320 Thermometer 425505 \- "Hortvet" Depression Difference, Depression Difference, O c. O c. C. O c. C. 0.4220 0.4192 0.0028 0.4100 0-0120 0.4643* 0.4496 0.0047 0.4409 0-0134 0.4870* 0.4809 0.006 1 0.4757 0.0133 0.0091 0*5200* 0.5144 0.0056 0.5109 0*5532* 0.5489 0.0043 0-5473 0.0059 0.5869* 0.5816 0.0053 0.5768 0.0101 0.6210 0.6163 0-0047 0.6090 0.0120 0.0105 depressions, found, (2) o- (3), found, M0-- (5)* Average difference . . . . 0.0048 * As calculated by Stubbs and E l ~ d o n . ~ (7) Differences between two thermometers, (3) - (5) 0.0092 0.0087 0.0052 0.0035 0.0016 0.0048 0.0073 In Table I1 are given the results obtained for the sucrose solutions normally used in the Each depression is the average of at least ten With one exception, where a difference between extremes standardisation of Hortvet thermometers.closely agreeing determinations. of 0-005" C. was recorded, the spread of results in each series did not exceed 0.003" C.254 SUTTON AND MARKLAND : THE STANDARDISATION OF [Vol. 75 It is apparent from the table that the depressions are not in agreement with Hortvet's figures for 7 and 10 per cent. sucrose solutions and the calculated depressions for solutions of intermediate strength. The differences are of the same order as those previously found with milk samples. The results with the two thermometers, however, are not in agreement in themselves.With thermometer No. 31320, the depressions recorded are on the average 0.005" C. less than those previously accepted for these solutions, and with thermometer No. 425505, the depressions are on the average 0.010" C. less. From column 7 in the table it is clear that there is reasonably good agreement between the two thermometers in the results for the 9 per cent. sucrose solution, but that differences reaching 0.009" C. are found in other parts of the scale examined. It may be suggested that the differences to which we have referred so far are entirely attributable to the difficulty in working with sucrose solutions. To test this point, portions of a large sample of milk were treated with amounts of water or strong lactose solution to give mixtures having freezing-point depressions near to those of the sucrose solutions normally used for standardisation.Freezingpoint determinations were made on each sample using each thermometer. Each depression in the table is the result of duplicate determinations. Usually these were in complete agreement but where slight differences were found an average figure has been recorded. The results are shown in Table 111. r Sample RESULTS WITH MILK (2) (3) (4) 7 Thermometer No. 31320 Freezingpoint depression, O C. - I Correc- Correc- tions tions from from sugar N.P.L. solu- certi- Difference, tions ficate (2) (3), C. (5) (6) (7) Thermometer No. 425505 r 1 Freezing-point depression, O C. Lorrec- Correc- tions tions from from sugar N.P.L.solu- certi- Difference, tions ficate (5) o- (6), -A C. (8) (9) Differences between results with two thermometers, C. & Correc- Correc- tions tions from from sugar N.P.L. solu- certi- tions, ficates, (2) - (5) (3) - (6) Milk + water . . 0.425 0.420 0.005 0.423 0.412 O.OX1 +0*002 +O.OOS Milk + water . . 0.456 0.451 0.005 0,457 0.444 0.013 -0.001 +0-007 Milk + water . . 0.489 0.483 0.006 0.489 0.477 0-012 0 + 0.006 Milk +water . . 0.620 0.514 0.006 0.520 0.511 0.009 0 + 0-003 Milk .. . . 0.546 0.540 0.006 0.546 0.541 0.005 0 - 0.00 1 Milk + lactose . . 0.588 0.583 0.005 0-588 0-578 0.010 0 + 0.005 Milk + lactose . . 0.622 0.617 0.005 0.624 0.612 0,012 -0.002 f0-005 The figures in this table show that, when standardisation with sucrose solutions is relied on, the results are in.excellent agreement (column 8).On the other hand, using corrections from the N.P.L. certificates, the depressions recorded are invariably smaller and in general by amounts similar to those recorded in the examination of sucrose solutions. There is, moreover, the same indication of differences in the results obtained with the two thermometers. Again the difference is small in measuring a depression of 0.55" C., but much greater differences are to be observed in other parts of the scale. A comparison of the differences in results obtained with these two thermometers following the use of corrections obtained from N.P.L. certificates is made in Table IV. The differences, both with sucrose solutions and milk, at any level of freezing-point depression are of the same order of magnitude and, since it is generally accepted that results with milk are easy to reproduce, it is suggested that the results with sucrose solutions are equally reliable. It is true that with these particular thermometers there is good agreement in the results for depressions near to those normally recorded for genuine milk, but much larger differences are to be observed in other parts of the temperature scale where presumably the same methods were adopted to ascertain corrections.CONCLUSIONS Our experience therefore does not enable 11s to support the proposals of Aschaffenburg and Hall. From our work with two thermometers there is no evidence that the use of correc- tions from N.P.L. certificates leads to more concordant results.Indeed, our results by thisMay, 19501 HORTVET THERMOMETERS TABLE IV 255 COMPARISON OF DIFFERENCES IN RESULTS OBTAINED WITH TWO THERMOMETERS AFTER CORRECTING THE OBSERVED READINGS FROM N.P.L. CERTIFICATES Difference in depressions recorded with thermometers Nos. 31320 and 425505 Results with No. 31320 - Results with No. 425505 Approximate freezing-point < A 3 depression, Sugar solutions, Milk, O c. O c. O c. 0-42 0.45 0.49 0.52 0.55 0.59 0.62 + 0.0092 + 0.0087 + 0-0062 + 0.0035 + 0.0016 + 0.0048 + 0.0073 + 0.008 + 0.007 + 0.006 + 0.003 - 0.001 + 0.005 + 0.006 procedure, within the range of temperature normally used, compare very unfavourably with those which are obtained with the same thermometers following the application of corrections based on standardisation with sucrose solutions.Further, if the proposed method were adopted, freezing-point depressions of milk would be at least 0405" C . smaller than hitherto recorded. It would therefore be necessary to revise the limit of 0.530" C. which has for so long been adopted as the smallest depression likely to be found for genuine milk. This, to say the least, would be unfortunate. Hortvet's figures for the depressions of 7 and 10 per cent. sucrose solutions may not have been strictly correct. Elsdon and Stubbs* report that they ascertained from the Bureau of Standards that the thermometers used would be correct to k0402" to f0.005" C. However, it has always seemed to us that the acceptance of Hortvet's figures represented a sound basis for the stan4ardisation of thennometers to be used in the Hortvet cryoscope, and it is true that all our records, some of them collected with considerable care, from which the limit of 0.530" C. has emerged, have been based on the adoption of these figures. It seems to us that it would be unfortunate to have to alter the limit of 0.530" C. solely to accommodate, for the standardisa- tion of the thermometer, a different system which apparently does not provide more urlifonn results. REFERENCES 1. Sutton, R. W., Markland, J., Barraclough, A,, and Chapman, W. B., Analyst, 1960, 75, 42. 2. Aschaffenberg, R., and Hall, J. -4., Jbid., 1949, 74, 380. 3. Stubbs, J. R., and Elsdon, G. D., ]bid., 1936, 61, 455. 4. Elsdon, G. D., and Stubbs, J. R., Ibid., 1934, 59, 685. ST. MARY'S GATE COUNTY OFFICES DERBY