of magnets in all plants, and in connection with their effectiveness when 

 installed in different mechanisms and under different operational methods « 



Test magnets o => This requires that each installation be tested to 

 determine its recovery efficiencyo Periodic tests of the same installations 

 ars made to discover fluctuations in performance and detect any changes 

 that may occur as plant operations are alteredo The test consists of 

 mixing a definite number of tagged dead fish with an ordinary lot destined 

 to go through the reduction system and recording the percentage of tags 

 recovered from the magneto A possibility that the efficiency of recovery 

 of test-tags may differ from that of tags in sea-released fish arises 

 from the fact that the former lie within the body, unattached to the 

 tissues, and not far from the open, insertion incision; whereas the iat=- 

 ter are encysted by tissue and the incision healed shuto It is likely, 

 therefore, that the efficiency may be somewhat over-rated by the testSo 

 On the other hand, a few recovery experiments indicate a lessened effi-- 

 ciency of recovery from fish destined for canning than from those des" 

 tined to go directly to the reduction machinerj^c This is probably due 

 to the loss of tags from the body cavity in the cutting and eviscerating 

 process. The performance of testing experiments in connection with canning 

 operations has met with obstacles preventing the accimulation of suffi- 

 cient experience to gauge the reliability of efficiency ratingso Their 

 value for adjustments is also lessened by the custom of mixing whole fish 

 with cutting offal in the ordinary operating procedure o In all, the ef- 

 ficiency of recover^'- varies widely and while important for our calculations 

 is difficult to determine o 



Analysis 



Adjust for tagging mortality o =■ Two basiis adjustments are made to 

 deduce the actual number of viable tagged fish releasedo The first util- 

 izes the live-car experience in adjusting release records downward by 

 the indicated amount of mortality induced by the tagging process, with 

 due regard to the size of the sardines o The second takes into account 

 the additional mortality brought about by the increased time of holding 

 to which the late-tagged members of a given batch of fish were subjectedo 

 It involves establishing a regression of returns on serial order of tagging 

 in blocks of 100 or other suitable number of individuals for each batch 

 or group of batches of fish taggedo These regression values adjust not 

 only for the time -connected differential mortality within a batch but 

 also for the differences in mortality between small batches tagged in a 

 short time and large batches requiring more time for tagging^ 



These adjustments have been employed in the analysis of retur^ns from 

 California-tagged fish, but no comparable methods are available for these 

 tagged in the Pacific Northwest, though it would appear that the second 

 adjustment at least would be feasible for releases in that areao 



