Nevertheless it’s important to notice that data could be updated 
and a careful and a careful terrain recognition could change the 
number of 2 values. 
  
  
  
  
Fig. 85 —The yellow colour is for structures (whose circular, square or triangular symbol shows classes: settlements, 
noble and ecclesiastic) lacking a precise date, other than the white structures bearing the related year, as reported by a 
foundation act. It is not a case that the last ones are found at lower levels. 
3. CONSISTENCY 
We have choosen to test a typical case of ambiguity of data 
recorded on old acts, which, in our mind, shows well the high 
grade of uncertainty to be found for identification of medieval 
structures with the use of anthropical elements of the same age. 
We deal with a grange having a toponym at present found at 
two different spots of the Po Valley: the toponym, in old papers, 
is assigned to two distinct areas and, in a third case (having a 
numerosity no lower than the other two ones), is reported 
without mention of area. 
We try, for assay purpose, the Bayes' Theorem, after translating 
some informations, available on ancient papers, into some 
indexes weighing upon paramount probability: 
_ P(A)x p(G/ A) 
PE Gt 
_ Pp(B)x p(G/B) 
EG p(Gtot) 
where: 
p(Gtot) — p(A)* p(G/ A)+ p(B)* p(G/ B) 
40 
Let's suppose that, on the basis of numerosity of documentary 
attributions to areas A or B, we assign to starting conditions 
p(A) and p(B) the values 0.6 and 0.4 [p(A)+p(B)=1]. Let’s 
further suppose to quantify as p(G/A) = 0.6 and p(G/B) = 0.35, 
other probabilities ensuing from the range of relateness of areas 
A and B to possible (and not uncommon) local morphological 
issues recorded by available acts (water, streams, woodland, 
routes). 
Conditional probabilities p(A/G) and p(B/G) are therefore equal 
to 0.72 and 0.28. 
Although the above reported example has been largely 
simplified for an easier treatment, we think that Bayes’ 
Theorem may be taken as suitable to solve ambiguities; of 
course, in case it should be used operatively for control and help 
for ambiguities, it would be quite necessary to find proper 
criteria for giving numerical values to probabilities. 
Solution of ambiguity problem also gives an improvement of 
location accuracy. 
For a historical-type GIS, a number of problems of spatial, 
thematical and time accuracy are present: for instance, it is often 
necessary to state the past position of an object whose traces 
were subsequently lost. The same is for a situation, an object, a 
facility (a mine, maybe) which has been alternatived active or 
not. 
The ba: 
sources 
the pos 
Fig. 87 
Fig. 88 
Castell 
Hidden 
A Hid 
parame 
un 
notice t