44 DAVID R. BRIGGS 



lion, while the osmotic pressure at 0°C. would amount to 51.4 cm. of 

 water. Absolute freezing point depressions can be measured, with 

 highest refinements of technique, to within an accuracy of perhaps 

 0.0025 °C. The vapor pressure of water at, say, 25 °C. cannot be 

 measured with an accuracy exceeding about 1 part in 10,000. The 

 osmotic pressure, however, can be estimated to perhaps 0.5-1.0 mm. 

 of water. It is obvious that with such a protein solution the freezing 

 point depression or the relative vapor pressure change would be un- 

 detectable but that, where direct measurement of osmotic pressure 

 can be accomplished, this method will yield results of a sufficiently 

 high order of accuracy to be usefid for the purpose of calculating the 

 molecular weight of the solute. (Boiling point elevation measure- 

 ments, while of about the same degree of accuracy as freezing point 

 depression measui'ements, are of little value for biological systems 

 and substances because of the many changes that may occur in such 

 systems when heated.) Because of the slight effect produced per 

 molecule on the relative vapor pressure or freezing point these 

 methods when applied to the determination of molecular weights of 

 solutes are most valualjle only when the molecular weight of solutes 

 is less than about 4000. 



While the classical works of Morse and Frazer {32,33) and of 

 Berkeley and Hartley {24) showed that osmotic pressure measure- 

 ments of high accuracy can be made on solutions containing low 

 molecular solutes, they also emphasize the great difficulty encountered 

 in preparing membranes semipermeable to such solutions. In gen- 

 eral, it is the conclusion of experimental studies that for solutes of 

 molecular weights less than about 10,000 to 15,000 it is very difficult 

 to prepare membranes sufficiently semipermeable to allow for de- 

 pendable measurements of osmotic pressure. On the other hand, 

 for a solute of molecular weight 500,000, dependable extrapolation of 

 data to an intercept value requires measurements at concentrations 

 for which the observed osmotic pressure is less than 0.5 mm. of water 

 and values so obtained become inaccurate because the probable error 

 of the measurement approaches or exceeds the actual measured 

 value. It is clear, therefore, that the usefulness of direct osmotic 

 pressure measurements in the determination of molecular weights 

 will be confined to those solutes having molecular weights between 

 10,000 and 500,000 and will be of true significance for this purpose 

 only if the solute is homogeneous. Molecular weights calculated 

 from osmotic pressure data obtained on a mixture of nondilTusible 



