g^ AI.AN ARTHUR BOYDEX. 



Tin- results shown in Table I. are (i) that readings of the ring 

 test generally corresponded at a given period, (2) that the titer 

 increased with time up to an hour, (3) that the greatest deviation 

 in the reading was plus or minus one tube or 50 per cent, in 

 terms of titcr, (4) that when the order of the tubes was that of 

 increasing dilutions and known to the reader a higher reading 

 was usually obtained than when the order was irregular and un- 

 known. This latter result is natural for if the tubes are arranged 

 in order of increasing dilutions of antigen, the rings occur strongly 

 in the lower dilutions and gradually less strongly in the higher 

 dilutions and the first tube to be apparently negative often on 

 close observation will be seen to possess a faint but definite ring. 

 This might easily be overlooked if the order of the tubes is un- 

 known. For this reason the order of the tubes has been known 

 in all titrations on relationships. 



Now a possible error of plus or minus 50 per cent, in reading 

 the ring test (which increases to 100 per cent, when a single 

 series of dilutions is used) doubtless seems very high but one 

 must remember that this is a biological test and that in spite of 

 this error proteins may be identified in dilutions far beyond 

 the range of any chemical method known at present. By 

 "possible error" is meant simply that inasmuch as the dilutions 

 tested were each 50 per cent, less than the preceding dilution, the 

 maximum error, should the end point actually fall somewhere 

 between two successive dilutions, would be 50 per cent. It is 

 to be remembered that this possible error is inherent only in the 

 serological titration not in the preparation of the antigen solution 

 (error less than 4 per cent, in nitrogen determinations). The 

 probable error would not be more than half as great as this 

 I>ossil)le error. It is felt that the use of a double series of dilutions 

 is justified on account of the more accurate reading thus ob- 

 tained. A triple or quadruple series of dilutions might be used 

 with an even greater accuracy but the labor involved in their 

 use is so great as to preclude their practice. 



2. The Sensitivity of the Ring Test. 



if tern antisera have been carefully titrated with standard 

 Their homologous tilers at one hour were distributed 

 as follows: 



