keys, (ibid, Stone). 
Although in some cases the views presented here appear to be 
in conflict with other viewpoints expressed in the literature, two 
general points of agreement may be identified in the field of image 
interpretation (using the terms given above): 
1. An interpreter with a low reference level will not be 
able to learn a great deal from his questioning of an 
image using an unstructured approach. 
2. An interpreter with a high reference level in a given 
subject may usually obtain a large amount of information 
with a high level of accuracy using an unstructured 
approach. 
It has also been stated that a key (or structured approach) 
is of limited use except for military interpretation (Vink, 1964) 
or to those with a very low reference level who are seeking rather 
readily indentifiable features. A concomitant opinion is that 
structured methods are of no use to expert interpreters or to those 
with a high reference level. Neither of these statements necessar 
ily follow from the numbered statements above, nor may they be 
supported in the light of the results of this research. 
It is the opinion of this author that the structured approach 
is very useful. This is particularly true where a large area is 
to be interpreted, the interpretation decisions tend to be repeti 
tive, and an individual with a high reference level cannot devote 
sufficient time to the project to ensure timely completion. 
Several steps which are necessary for successful use of the 
structured approach may be identified. It is assumed that only 
a scientist who has considerable experience in both image inter 
pretation and an appropriate subject area would attempt to follow 
the steps given below. 
THE STRUCTURED APPROACH 
1. Assess the system under study. Answer two questions: 
a. Can the feature of interest be directly recognized on 
the image by non-experts? If yes go to 2.; if no go 
to l.b. below. 
b. Does the feature have an image surrogate of high 
reliability? (Where an image surrogate is something 
that may be recognized directly on the image by non 
experts and which indicates with acceptable relia 
bility the presence of the feature of interest.) 
If no surrogate is present consider other imagery 
scales and types and begin l.b again; or reject the