198 REV. S. M. JOHNSTON 



where C is the value of the so-called boiling-point elevation constant expressed per 

 gramme particle (molecule or ion) in one gramme of solvent — 



wi = molecular weight of salt added. 

 \Y = weight of solvent used in grammes. 

 w = weight of salt added in grammes. 

 a = ionization coefficient. 



n = number of free ions into which a molecule of salt dissociates. 

 E = elevation of boiling point. 



From this formula values of C were obtained, at one time high, at another low, when 

 compared with theory. Thus for potassium chloride the values 858, 704, 684, 643, 

 614, 596, 572 were obtained from one series; a second series gave 460, 467, 514, 

 518, 523. For potassium nitrate the values were 637, 617, 603, 573, 571, 549, 547, 

 540, and a second series gave 874, 702, 696, 686, 643, 608, 609, 593 ; for a third 

 series the values were 605, 556, 550. For sodium nitrate 518, 516, 520, 529, 534, 

 530, 535. For sodium chloride 617, 621, 592, 587, 579. 



Such values as these being obtained, it was desirable to see what values would be 

 given by calculations from the boiling-point data of other experimenters. Elevations 

 of boiling point as given by Biltz # gave for potassium nitrate as values 638, 589, 

 596, 628 ; and for sodium chloride 585, 598, 611, 609. 



Those given by Walker and LuMSDENt gave for potassium nitrate 648, 618 ; for 

 sodium chloride 598, 593 ; and for potassium bromide 620, 614, 645, 665. 



Smitz'sJ elevations gave for sodium chloride 463 ; for potassium chloride 497 ; for 

 potassium nitrate 522 ; and for sodium nitrate 594. 



In discussing the meaning of the high and variable values of the elevation constant 

 which have just been given, I have considered (l) whether they are due to error in 

 correcting for change of atmospheric pressure, (2) to overheating. 



(1) To test the values obtained under the first supposition I set up two similar 

 boiling-point apparatus. The gas pressure was equalised by the use of a three-way 

 tube, and two Bunsen burners exactly the same were used. The strength of the source 

 of heat was therefore the same for each tube. The quantity of solvent used for each 

 was 25 cubic centimetres. The one tube was used for solutions, the other for solvent. 

 The readings in each instance were taken from Beckmann thermometers reading to 

 one-hundredth of a degree. In this way it was possible to obtain a thermal register 

 of the change of atmospheric pressure during the observation of boiling-point elevations. 



For a series of experiments, readings were made of the barometer, the boiling point 

 of the solvent, and that of solutions of sodium bromide. The observed boiling tempera- 

 tures of solutions were then corrected by the readings of the barometer, and also by the 

 thermometer in the tube containing the solvent. 



'& 



* Practical Methods for determining Molecular Weights, translated by Jones and King, p. 189. 



t Journal of the Chemical Society, 73, 502 (1898). J Zeit. fur phys. Chemie, 39, 420. 



