February 27, 1896] 



NA TURE 



389 



In his last letter, Sir Henry Howorth expresses his belief that, 



- a consequence of the recent correspondence, I have been led 



. change the opinions I previously held as to the cause of an 



ice Age. May I assure Sir Henry Howorth that he must have 



quite mistaken the purport of my letter. 



With all due courtesy to Sir Henry Howorth, as well as to 

 any others who may have differed from me, I may say that 

 I have seen no reason to swerve from the belief that the 

 ])()sition taken in my book on "The Cause of an Ice Age" is 

 the sound position. My opinions are, therefore, unchanged. 



I would like to take this opportunity to thank Dr. A. R. 

 Wallace for his letter on this subject in your columns, and to 

 express the satisfaction with which I found that he had been led 

 to the same conclusion, with respect to Mr. Culverwell's argu- 

 ment, as that at which I had myself arrived by an independent 

 method. I note that Mr. Culverwell dissents ; but. even at the 

 risk of being tl^ought very obstinate, I must say that I still believe 

 Dr. Wallace and I are right. Robkrt S. Bali.. 



.So far as I know, it has not been suggested that a compara- 

 tively small elevation of the crust of the earth would cause a 

 glacial state in the elevated part. The roughness of an ordinary 

 terrestrial globe, representing an elevation that might be com- 

 pared with the thickness of letter-paper, would correspond with 

 an elevation of the crust of the earth that even in the tropics 

 would cause perpetual snow. A few thousand years would be 

 -ufficient to cause a sufficient elevation without any catastrophic 

 hypothesis. Theodore Rylan'd. 



The Measurement of High Temperatures. 



In a valuable paper on the " Determination of High Tem- 

 peratures" ( ^^''<?'^. Ann. 1895, No. lo), Messrs. Holborn and 

 Wien give the results of their observations on the changes in the 

 lesistance of platinum wire over a range of o° to 1600° C. The 

 authors come to the conclusion that the relation between 

 temperature and resistance "cannot be accurately represented 

 l)y the Callendar and Griffiths formula"; although, on the other 

 hand, they admit that by means of that formula Heycock and 

 Xeville have determined a nuinber of melting points which are 

 in good agreement ("diesich in guter Uebereinstimmung mit 

 unsern Werthen befinden") with the values found by Holborn 

 and Wien when using a thermo-couple standardised by direct 

 comparison with the air-thermometer. 



Platinum thermometers are now so generally used for high 

 temperature measurements that an adverse conclusion of this 

 kind is a matter of importance, especially when associated with 

 the names of accurate observers such as Holborn and Wien. 



I trust, therefore, that a brief criticism will not be regarded as 

 out of place. 



I would first remark that my delay in commenting on this 

 paper is not due to any want of respect for the authors, but from 

 a sense of the importance of the matter. Before attempting any 

 reply, I wished to ascertain the views of Prof. Callendar, and I 

 accordingly wrote to Montreal calling his attention to the paper, 

 but I have not as yet received his answer. I feel, however, 

 1 hat further delay is undesirable. 



Neither Prof. Callendar nor I have at any time claimed 

 that the relation between t and //, as given by the empirical 

 formula, has been rigorously verified at temperatures exceeding 

 600° C. 



In Nature, November 1895, P- 40> ^ wrote as follows : — 

 " Results of this kind prove that even if the reduction does not 

 express the temjjerature accurately in the air-scale, it at all 

 events gives us a constant scale in which all high temperatures 

 can be expressed, and it is further evident that this constant 

 scale differs but little (even at these high temperatures) from the 

 true air-scale." 



The context shows that the "high temperatures" referred to 

 were those in the neighbourhood of 1 100° C. 



The above quotation defines my own position with sufficient 

 accuracy, and I will therefore pass on to consider the work of 

 Messrs. Holborn and Wien. 



( I ). The authors state that the platinum wires (in experiments up 

 to 1 200° C. ) were placed in an externally glazed porcelain tube, and 

 isolated from each other by means of special porcelain capillaries 

 ("die DrJihte wurden durch besondere Porzellan Capillaren von 

 einander isolirt " ) ; and at higher temperatures "in externally 

 glazed tubes of a very infusible substance, and isolated from 

 each other by capillaries of the same material." 



NO. 1374. VOL. 53] 



From this description it appears evident that the wire was in 

 contact with the porcelain, or the other material, probably 

 throughout the greater portion of its length ; otherwise it must 

 have been subjected to some strain. Now, contact with porcelain 

 or any similar substance has been found by us to be absolutely 

 fatal : if by any chance the wire has come into contact with the 

 walls of the surrounding tube at high temperatures, the coil has 

 had to be replaced by a new one. This is probably caused by 

 the action of the silica ; but whatever the reason may be, the 

 eft'ect has long been known, and care has been taken to avoid 

 this source of error. 



It is true that in the present form of the platinum thermometer 

 the wire comes into contact with mica, and I have little doubt 

 that some similar action (although in a lesser degree) takes place 

 between the platinum and the mica. The method of construct- 

 ing the framework and coil, however, causes the length oi 

 platinum in actual contact with the mica to be but an exceed- 

 ingly small fraction of the length of the wire (probably less than 

 li^th), and thus any such effect is diminished. I think it 

 possible that the small changes at high temperatures to which I 

 have called attention in Nature, (November 1895, P- 4°)' are in 

 some measure due to action at the points of contact. 



Unless Holborn and Wien took greater precautions in regard 

 to this matter of contact between case and wire than they indi- 

 cate in their paper, there is no need to seek for further explana- 

 tion of the somewhat erratic behaviour of the two platinum wires 

 examined by them ; in fact, other experiments by the same 

 observers show how materially the resistance of platinum is 

 affected by exposure to the action of silica or hydrogen. 



(2). No adverse conclusion should be drawn from the changes 

 in their Wire No. I., for the authors state that " at the termina- 

 tion of the observations the protecting tube was found to have 

 broken in the oven, and the wire had been exposed to the gases 

 of the oven." 



The necessity of complete protection from the furnace gases 

 has from the first been insisted upon by those accustomed to the 

 use of platinum thermometers. It was want of attention to this 

 essential matter which led the B. A. Committee of 1874 to a 

 false conclusion. The behaviour of Holborn and Wien's Wire 

 I. has no significance or value, except in so far as it emphasises 

 the importance of complete protection. It appears doubtful if 

 sufficient precautions were taken in this matter with regard to 

 their series of experiments with the platinum Wire II. The in- 

 fusible tubes which they used for the air-thermometer bulbs at 

 high temperatures were covered externally with a glaze. The 

 authors say : " As the glaze in this case and also in the porcelain 

 becomes liquid much sooner than the softening point of the 

 material, we made use of our method by which there must 

 always be a smaller pressure inside the vessel than outside. 

 Under these conditions the glaze is pressed into the pores of the 

 tubes, otherwise it would immediately come off." 



Now I do not find any mention of a similar precaution when 

 heating the platinum wire in what appear to have been similar 

 tubes ; it is possible, therefore, that the changes observed in 

 Wire II. were also in some degree due to contamination by 

 furnace gases. 



(3). The authors speak of Wire II. as formed from " pure 

 platinum" {"Aus reinem Platin "). They give the value of its 

 temperature coefficient as — 



Before heating 

 After 



•003801 

 •003783 



As a general rule, the purer the platinum the higher its co- 

 efficient. The samples used by us (as, for example, in the Kew 

 Observatory thermometers) have coefficients which vary from 

 •003860 to -003880. It would hardly appear, therefore, that the 

 wire used by llolborn and Wien merits the term " pure." The 

 purity is, however, not of great consequence, as (see Nature, 

 November 1895, p. 40) we have found that although the co- 

 efficient depends on the purity, the deduced temperatures are 

 unaffected provided that the coefficient is not reduced by the 

 impurities to lower than about ^0032. The fall in the coefficient 

 (above indicated) is, however, of great .significance, and in itself 

 is evidence that the wire had become contaminated during the 

 experiments. Assuming (as was doubtless the case) that the 

 wire had been previously annealed, the above change sufficiently 

 establishes inadequate protection of the wire. 



I have carefully investigated the numbers in Table II., where 

 the authors give the results of their observations on this second 

 specimen of "pure platinum." I am unable to draw any con- 



