Sec. 8.7] INDIRECT METHODS FOR MEASURING DEUTERIUM 273 



centrations of to 3 per cent at 26.8°C and which avoids the differential 

 vaporization of a binary system. w-Fluorotoluene may be used for the 

 same concentration range at 19.3°C. These toluenes have the further 

 advantage of lower viscosity and therefore greater sensitivity. They may be 

 obtained from the Eastman Kodak Company or synthesized by a procedure 

 analogous to that for fluorobenzene [49]. 



For higher concentrations of deuterium oxide, o-fluorotoluene is unsatis- 

 factory because the precision of the method falls off rapidly as the difference 

 between the density of the water and the reference liquid increases. For 

 samples containing 10 to 40 per cent deuter um oxide, solutions of phenan- 

 threne in a-methylnaphtha ene have been shown to be superior reference 

 liquids [50]. Neither of these components is particularly volatile. The 

 exact viscosity of o-fluorotoluene has not been published, but it is estimated 

 that that of a-methylnaphthalene is about three times as large; so for the 

 same density difference, the drops fall more slowly in the latter medium. 



The dimensions of the tube containing the medium are not critical except 

 that the inside diameter should be greater than three times the diameter of 

 the drop to avoid wall effects. The fiducial marks should be as far apart 

 as possible (about 20 cm) to provide maximum sensitivity and the best 

 average rate of fall. Another 20 cm should be allowed above the top mark 

 to ensure thermal equilibrium between the drop and the medium by the time 

 the drop reaches the first mark, and about 10 cm should be allowed below 

 the lower mark to avoid end effects. 



The size of the drop is controlled through the use of a mechanical micro- 

 pipette. The size is not critical and may be 5 to 45 mm 3 [54], but it is impera- 

 tive that it be uniform in any set of determinations, including the associated 

 calibration. It is desirable to use as small a drop as possible to minimize the 

 quantity of water needed for an analysis, but the sensitivity of the method 

 increases with the square of the radius [54]. Variations in drop size introduce 

 a percentage error in velocity which is two-thirds the percentage error in 

 volume. For falling velocities of the order of 1 cm per sec, the variation in 

 drop size must be less than 0.1 per cent. 



The designs for several satisfactory micropipettes have been published 

 [14,55-58]. The accompanying drawing (Fig. 58) illustrates the one used 

 by the authors [59]. It is somewhat less bulky than other published instru- 

 ments but provides the required precision. The barrel K is made from a 

 34-cc glass tuberculin syringe and is joined to the capillary with cobalt glass. 

 Mercury is displaced by the advancement of a piece of K6 _ i n - diameter drill 

 rod driven by a micrometer head. A mechanical stop A assures uniformity 

 of the angle through which the micrometer is turned. With the sample 

 in the capillary, the tip is lowered about 1 cm below the surface of the medium 

 and a drop formed. The drop is released by raising the pipette slowly and 



