TM No, 377 



This bias is probably felt most in the covariance function* Even if the 

 biased regions are randomly dispersed throughout a record pair, the spurious 

 d-c effects (because of correlation) would not necessarily tend to average 

 out. Further discussion of the biasing problem is best deferred until the 

 examination of some actual field data. 



Errors in Bookkeepi ng — Other sources of minor errors included improper 

 tape labeling, accidental switching of title format cards, etc. — errors of 

 bookkeeping for the great number of data samples, card decks, and the like= 

 Minimizing these errors requires careful and systematic procedures, backed up 

 by the drudgery of checking and double -checking all data at each step of the 

 reduction and processing. 



70 



