CIP A 2005 XX International Symposium, 26 September - 01 October, 2005, Torino, Italy, 
650 
work, needed for the subdivision of the building blocks into 
cadastral units, for the relief of the courtyards, etc. 
In such a way, moving backwards over time trying to find 
information about the techniques and methods of survey of the 
Census Map of 1866, we ended up by finding about the rigorous 
relief, carried out with scientific topographical criteria, of the 
Map of Rome by Nolli, of 1748. The long work of Nolli - lasted 
twelve years - is characterized by the exceptionally high level 
of accuracy and detail. Its previous working experiences in the 
cadastral operations of Lombardy and Savoia had contributed to 
its technical formation, also as far as it concerns the use of the 
tavoletta pretoriana, employed during the survey (see Figure 1). 
Perfectly aware that the tavoletta pretoriana, in spite of its high 
precision, could not guarantee a result "of the reduced margins 
of approximation" (Bevilacqua, 1998), Nolli carried out a 
triangulation of the entire city within the Aurelian Walls, based 
on check points situated on columns and obelisks, and along the 
axis of Via del Corso. Consequently, the survey of the single 
city sectors could be done, starting from the main public 
squares, streets and monuments, and sharing the eventual 
discrepancies over less important built up areas. 
Figure 1. The tavoletta pretoriana and its use (from Marinoni 
1751). 
Coming back to the Topographical Plan of Rome of 1866, by 
the Direzione Generale de! Censo , it can be considered as an 
up-to-date version of previous cadastral cartographies which 
seem to be based on the Nolli’s cartography. It appears, 
according to this hypothesis, that Nolli would have been the 
only one who caried out a topographical survey of Rome, long 
ago, in the middle of the XVIII lh century. Keeping in mind the 
difficulties of comparing such dated products with 
contemporary cartographies - created with different techniques - 
the necessity to analyse the urban trasformations, imposes the 
study of appropriate methodologies for referencing the 
historical cartography in actual coordinate systems. 
2.2 The georeferencing procedure 
During the first phase of the project, the georeferencing of the 
Census Map (1866) was carried out, using as reference the 
coordinates of the Regional Technical Map of Lazio (1990) in 
scale 1:10.000, which was at the moment the only official 
cartography available. The procedure of “tying” parts of 
historical cartography to modem cartography allows a local 
esteem of the error, guaranteeing a control over the 
georeferencing procedure accuracy, necessary for the eventual 
redefinition of the control points. The overlaying of the 
georeferenced Census Map with actual cartographies has shown 
* From now on “The Census Map” 
systematic errors occurring in some areas. The major 
discrepancies remain however within the medium error level, 
that is of approximately 4 meters, suggesting the existence of 
probable historical errors, caused by wrong measurements 
propagated during the survey network. The cartographic error, 
related to numerous causes, has been subject of a detailed 
examination, in order to define the influence on the accuracy of 
the final product. The main causes of error have been identified 
as follows: (i) deformations of the historical cartography, due to 
problems of conservation of the original support and probably 
to the different projection; (ii) historical error of survey, 
meaning propagation of rough errors along the survey network; 
(iii) historical error of representation, related to the presence of 
nonexistent particulars, added to the map for aesthetical 
reasons; (iv) contemporary error of interpretation, related to the 
lack of detailed and uniformly distributed information regarding 
morphologic features, to be considered as unchanged between 
two successive cartographies; (v) contemporary cartographic 
error (see Fig. 2), related to inaccuracy in the contemporary 
cartography to be used as basis for georeferencing the historical 
cartography. 
Figure 2.. Contemporary cartographic error. Aurelian walls (San 
Giovanni). On the left, aerial photograph showing the real trace 
of the walls (white line); on the right, the Regional Technical 
Map of Lazio, showing a wrong trace. 
As already pointed out, georeferencing an historical map - 
which lacks of reference system - can only be done by 
calculating the coordinates of well identified cartographic 
elements. The fastest way to determine these coordinates is by 
retrieving them from actual cartographies on which the same 
elements are visible. It is clear that for a correct procedure, the 
actual cartography should be of a greater scale than the 
historical one, in order to avoid “forcing it” to a less accurate 
geometry. Since such a cartography at the time was not 
available, and the comparison with actual cartographies at 
smaller scale had led to contradictory results, we decided to 
proceed with a differential GPS survey of the entire city centre, 
for calculating the position of the identified cartographic 
elements in an absolute coordinate system. 
The survey, initially imagined exclusively for controlling 
purposes, was planned to homogenously cover the study area. In 
the first stage the localization of “correctly measurable” points 
on the historical cartography was carried out. The most suitable 
points are those incorporated within building bodies (for 
example building corners). Less suitable are to be considered 
easily recognisable points such as fountains or obelisks, that 
could have encountered modifications in the course of the years, 
due to urban rearrangements or street surface repairs. 
Unfortunately, the coordinates of the building comers are 
difficult to measure with GPS receivers, since the positioning 
into their base does not guarantee a good GPS cover (great part 
of the sky is invisible), while reaching a wall or a building 
comer at the top, so as to assure a complete GPS cover, involves 
an increase of logistic access problems. The first GPS network 
comprised approximately twenty suitable points, distributed as 
uniformly as possible over the city centre. This enabled the 
evaluation of the statistical shifting parameters. The geometric