for adults is cari'ied on only from late July or Aunjust to February or March, 

 and fish in sufficient quantity for study are not available during the inter- 

 vening time, it is not feasible to follow an entire year's cycle for adults 

 as it was for the young. In July, however, v;hen specimens were first taken, 

 the last ring was found to be fully formed in all exarained, and was completely 

 surrounded by a narrow marginal growth zone; -and no specimens were found 

 with developing new rings. During succeeding months the margin beyond the 

 las f ring seemed to increase in width, and by late fall nev; rings began to 

 appear in a few specimens. Iiiliile these were in the process of formation, 

 they could be easily distinguished from old rings by the follo'v.'ing criteria: 



an annulus is recognized as forming v/hen it does not appear entirely 

 around the sculptured part at once, but generally only on the most distal 

 parts of the edge (see figure 3)j or, when it extends entirely around the 

 sculptured part, it touches the edge at one or more points. 



These criteria seemed to serve for distinguishing new rings -until about 

 early January. From then on, however, an increasing number of specimens 

 occurred vfith the last (subraarg:'jial) ring remarkably close to the edge of 

 the scale, but complete, and entirely bounded by a marginal zone. For these 

 the above criteria could not be applied. Instead, at the time of first 

 reading, the operators tried to judge whether they were new by v/hether or 

 not they v;ere at any point of their course obviously closer to the edge 

 than could be accounted for by a year's growth. Unfortunately, there are 

 intermediate cases v«-hich could be attributed equally vrell to a year of ab- 

 normally slow growth or to a fe-v; weeks of abnormally rapid grovrth. Conse- 

 quently, the decisions on these cases rested more or less with chance, and 

 as shorm by the studies on parallel readings, this was one of the chief 

 causes of disagreemem; among the readers. 



In an effort to devise an objective standard for distinguishing new 

 from old rings, therefore, a statistical study v/as made on the width of 

 the marginal growth zone. One "vvould expect the latter to be narrowest shortly 

 after a new ring has formed | an^ to increase thereafter in width during 

 the year, reaching its maximum just before the follovdjig new ring appears 

 (Walford and Mosher, 19U3)« Accordingly, tho distance between the last 

 annulus' and the edge of the scale — the marginal zone — ^was measured to the 

 nearest 0.2 millimeter of projected image (equivalent to 0.0066? millimeters 

 in actual width), on all scales taken from July I9J4O through February I9UI. 



These- data wjre tabulated by number of rings ( not by assigned year, 

 classes), according to width of margins, in the form of a frequency distri- 

 bution. For example, all scales having fojir rings wore tabulated together. 

 It ivould be expected that early in the season, they would all be of year 

 class 1936; later in the season some of them (those with new rings) v/ould 

 bo of year class 1937 • The tv^ groups vrauld be expected to have the same 

 relative variation about their respective moans, and to put this on an 

 absolute basis for purposes of graphical study, sizes v/ere converted into 

 logarithms, which were tabulated into classes of equal intervals (Schrek 

 and Lipson, 19Ul). In figure ^> the results of this transformation are given 

 for scales with four rings, as an example of a case of average difficulty. 

 A dominant mode to the right, v/hich can be follo^/cd through the season, is 

 taKen to represent the margins attributable to the past season's growth; 

 and a nuw mode to the left, appearing in November, probably represents margins 

 attributable to the next season's grovfth. 



106 



