Coding symbols may be printed directly on cards, or blank cards 
may be used with a "key" card showing the code. It is more convenient, 
caches tnd costs little more to have all cards printed with the code 
symbols, 
The body of the card is available for recording any desired data. 
For example, a bibliography card file may have such items as subject, 
journal, date, author, species and file number coded in the notch 
positions, with the complete bibliographic citation, including notes or 
abstracts written on the body of the card. For a research project, 
the code may cover species, locality, date, observer, etc., with blanks 
in the body for writing in observations such as weights, frequencies of 
activities, and the like. 
Aside from a few general principles such as those described, coding 
is largely a matter of imagination and personal resourcefulness in de- 
vising code systems. The punch-card companies and their distributors 
offer free counseling services to help with refinements of all phases of 
the technique. 
Limits of the Punch-card Systems 
' The question may arise as to the advisability of using a punch- 
card system for a particular job. It is difficult to zeneralize about 
this, but it may be pointed cut that there are, perhaps, limits beyond 
which these systems should not be used. If the data are few, it may 
be just as quick to work them up manually. It is hard to say where 
the minimum limit is. Five hundred cards are probably near the minimun 
for the notch-card system. However, Cottam and Curtis (1948) used only 
100 cards with the Hollerith system which is generally considered to have 
@ much higher minimum efficiency number than the notch-card system. 
There is hardly any limit to the maximum number of cards used. With the 
Hollerith system, the more cards there are the more efficiently they 
can be sorted. This is not so true of the notch-card system. But at 
just what point of numbers the Hollerith system becomes more efficient 
than the notch-cards cannot be stated since it would vary under differ- 
ent conditions. In addition to number of cards, efficiency is determined 
in part by curd code capacity and number of times they are to be sorted. 
A little experimenting with the punch-card systems will help to determine 
individual limitations. 
Supplementary Equipment 
Both systems have supplementary machinery to suit various purposes. 
Several thousands of dollars are needed to buy a complete outfit of 
Hollerith machines. These machines will perform unbelievably complex. 
operations such as mathematical computations and transforming coded 
material into printed reports. The notch-card system supplementary 
