320 



NA TURE 



L/«fy 



gether disagreeable. Fig. 5 shows exactly the stiff" attitude 

 the head is compelled to take in order to rest. 



Finally, Fig. 6 reproduces the comfortable position in- 

 dicated in Fig. 3, and at the same time represents the 

 profile of the back of the seat actually in use in our rail- 

 way carriages. On comparing this profile with the posi- 

 tion of the man comfortably supported, the following 

 defects in the back of the seat are observed : — 



1. It is too vertical. 



2. It allows an empty space between the lumbar verte- 

 bra? and the lower extremity of the shoulder-blade exactly 

 at the place where one is in the habit [of putting a 

 cushion " behind the back," as it is called. 



3. It is at least half a foot too high, and so makes it 

 impossible for the head to rest behind. It is customary 

 to make the back of the seat tally with the height of a 

 man of average size seated bolt upright. 



Under the actual conditions, such as they have been 



described, what becomes of the traveller when sleep at 

 length overtakes him ? Little by little he slides down on 

 his seat till the lower extremity of his shoulder-blades, 

 which has most need of support, finds the most sensible 

 projection, which, as the backs of our railway carriages 

 actually are, is precisely where it is least serviceable — at 

 a point, namely, on a level with the region of the pelvis. 

 Lastly, the head inclines forward or to the side, if it does 

 not bury itself in the breast (Fig. 7). 



Fig. 8 gives a front view of the face of the bench serv- 

 ing as the back of the seat. In the centre is seen a 

 stuffed projection, on each side of which a passenger 

 may rest his cheek. The shoulder, getting no separate 

 support, must contrive to lodge itself between this stuffed 

 projection and a kind of plateau fixed in the side of the 

 back of the seat, and which, situated about a hand's 

 breadth above the seat, offers a resting-place to the elbow 

 (Figs. 8 and 9). 



A NEW PRINCIPLE OF MEASURING HEAT 

 THE following method is intended to fulfil some con- 



*■ ditions which probably will be more and more 

 urgently required in the progress of modern science : — 



I. Measurements of heat should be executed at constant 

 temperature, i.e. without the aid of thermometers. — Every 

 variation of temperature during calorimetric experiments 

 causes unavoidable errors and necessitates corrections and 

 compensations. The accuracy of the thermometric method 

 [" method of mixture," of Regnault], which now predominates 

 among experimentalists is unrivalled, only in those cases 

 where the amount of heat to be measured is developed in 

 the course of a few seconds or minutes ; it is seriously im- 

 paired whenever the experiment lasts longer, while the 



influence of the corrections for radiation, &c, increases 

 proportionately with the duration of the operation. The 

 first method used in thermo-chemical investigations, the 

 ice-melting method of Lavoisier and Laplace, as well as 

 the modern calorimetric method by Bunsen, avoids this 

 inconvenience by executing all measurements at the 

 melting-point of ice. Bunsen's ice calorimeter is, how- 

 ever, not exempt from corrections. Every physicist familiar 

 with the use of this instrument will also, like the author, 

 be well acquainted with its capriciousness. Bunsen pre- 

 scribes that the calorimeter should be placed in a large 

 vessel filled with absolutely pure snow. Although I have 

 had abundant quantities of the purest snow at my disposal, 

 I do not hesitate to declare, having tried, in company with 

 Prof. Nilson, a whole winter to obtain reliable results with 

 the original arrangement of the inventor, that the instru- 

 ment would be impracticable for use without the improve- 

 ment devised by Schuller and Wartha, viz. to immerse the 

 calorimeter in a vessel containing ice and pure water at 

 o° C. Still the advantage of this arrangement is not to 

 prevent variations in the position of the mercury index, 

 but to make them quite regular. These variations are 

 declared by some physicists to depend upon the vacil- 

 lation of atmospheric pressure, but I think that the real 



Fig. 1 



cause of the unsteadiness of the index of the instrument 

 lies in the peculiar behaviour of the ice in the vicinity of 

 its melting-point. It was believed hitherto (and Bunsen's 

 method theoretically rests upon this assumption) that ice 

 at 0° C. suddenly changes its specific volume from that of 

 ice [= ro9o6S6] ' to that of water [= roooooo]. I admit 

 that this assumption may be true with regard to abso- 

 lutely pure iee, but in every kind of frozen water which 

 contains the smallest trace of impurity (which is unavoid- 

 able if the water has been boiled assiduously in a glass 

 vessel) the transition of ice into liquid water is not sudden, 

 but gradual, and begins already a III tie below o° C. Such 

 ice does not attain its maximum of volume exactly at o° C, 

 but some hundredths or tenths of a Centigrade degree 

 below o c (dependent upon its relative purity). Graphic 

 representation 2 shows that the co-ordinate of specific 

 volume of the ice comes not to a point d' arret at zero, 

 but moves upon the rapidly-sloping branch of a curve just 

 in the immediate vicinity of the melting-point. Now sup- 

 pose the water in the external vessel to be either a little 

 purer than that of the calorimeter, or vice versd. 



1 This number, which is almost identical with that of Bunsen, 

 by the author in his research "Upon the Properties of Water 

 " ^VfYi-expeditionens vet. iakttagels," Ed. ii. p. 275. 



- See the paper " Upon Water and Ice," by O. Pettersson, I.e. 



In the 



