GVII-74 
PHOTOGRAMMETRIC ENGINEERING 
tion are more accurate than boundary lines located by orthodox field checking 
methods. In this particular study on map accuracy there were only two in 
stances out of 77 field checks where the actual soil boundary was more than 100 
feet removed from the location on the map. It is well to point out that field 
checking the finished map is a combined check on the photo interpretation, the 
transfer of detail to the base map, and on any possible drafting errors. Thus an 
error of 100 feet might actually be attributable to transfer of detail by the 
sketchmaster or to drafting error rather than to photo interpretation. 
Had there been more errors between the photo interpretation and the field 
check of the boundary line location, then the errors would have been sorted 
Table III 
Hypothetical Data to Illustrate Analytical Methods for Determination of 
Boundary Line Accuracy between Map Units 
Accuracy Increments 
Adjoining Types 
of 100 feet 
R/BG 
R/B 
R/G 
R/F 
R/S 
R/P 
total 
0 
10 
7 
3 
2 
2 
1 
25 
1 
3 
2 
1 
0 
0 
0 
6 
2 
2 
0 
0 
0 
0 
0 
2 
3 
1 
0 
0 
0 
0 
0 
1 
Total 
16 
9 
4 
2 
2 
1 
34 
Analysis of Variance 
Due to 
D.F. 
Su 
m of Squares 
Mean 
Square 
F. 
Ratio 
Between accuracy Units 
3 
72.47 
24 
.16 
7 
.09* 
Within accuracy Units 
22 
75.00 
3 
.41 
Total 
25 
147.47 
Correction factor 
1 
38.53 
Grand Total 
26 
186.00 
* Significant at 99:1 odds. Of the 3 degrees of freedom the independent comparison between 
(0) and (1, 2, and 3) accuracy units accounts for 69.0 of the 72.47 in the sum of squares. The other 
two independent comparisons are not significant. 
The interpretation of the above hypothetical analysis is that errors of 100 feet or more in 
delineation of boundary units will occur less frequently than zero errors. 
and analyzed as in the hypothetical illustration in Table III. Snedecor (9) dis 
cusses the analysis of variance technique in chapters 10 and 11 appropriate to 
such an analysis. No tables similar to those for the binomial theorem are avail 
able thereby making necessary the computations for each set of data. 
Results 
A comparison of Tables I and II shows that the number of areas in the field 
check was not proportional to the number of map unit areas because some of 
the larger map unit areas were sampled in several places. Tables IV and V 
present the results of the statistical analysis of each soil type or map unit. 
The R soil type is used to illustrate proper interpretation of the tables men 
tioned above. Seventy-three soil type areas, out of a total of 493, classified as