nitrogen factors for the individual fractions as well as factors lumped to conform with the five fractions obtained by free-solution or paper electrophoresis. While the factors for the composite fiactions are reasonably close t o one another, it is evident that very large errors would arise if a serum containing abnormal amounts of some subfraction Jvith a widely d i v e i g ~ n tnitrogen factor were analyzed. The large va.iations in nitrogen factors which are indicated by the standard deviations in Table I11 reflect the wide differences in the composition of corresponding fractions from different individuals. These variations arise in part from variations in the amount of nonpeptide material, such as lipides, associated with the migrating fractions. A much more important cause of variations is that the 11 fractions which are discussed are mixtures
rather than pure proteins, so that their nitrogen factors vary, depending on which component predominates. This complexity is easily demonstrated by excising fractions from a completed electrophorograrn, reimplanting them, and continuing electrophoresis. After runs totaling 80 em. in length, a t least 14 fractions can be distinguished. This technique can be used to extend the effective length of an electrophoretic migration; its usefulness is limited only because progressive attenuation eventually makes the protein precipitates impossible to see. Fractions in an untreated electrophorogram can be located by comparison with a second electrophorogram which has been run in parallel, then treated with trichloroacetic acid. There is additional evidence for the heterogeneity of electrophoretic fractions in the irnmunoelectrophoretic studies of Williams and
Grabar ( 5 ) , during which 16 or more antigens were found in human serum. LITERATURE CITED
(1) Gordon, A. H., Keil, B., Sebesta, I