estimation was made using a Silicon Graphics VGX 
computer with 32MB of RAM. Some statistics relating to 
this rigorous least squares estimation are shown in Table 
1. For about 40 measurements, the least squares residuals 
were larger than three-times their a priori values, giving 
an initial a posteriori variance factor of 1.4 (significantly 
greater than unity). Those 40 measurements were 
"de-weighted" by a combination of the Danish’ method 
of robustified least squares and the judgement of those 
who carried out the measurements (survey and 
photogrammetric). In all cases the measurements were to 
targets only 1 or 2 meters from the instrument (camera or 
tacheometer) so the a priori standard deviations were too 
low. After this process, the variance factor was 
insignificantly different from unity (5% significance) and 
the a posteriori variances shown in Table 1 indicate 
homogeneous precision, apart from 11 poorly-defined 
natural features (pass points), each of which appears on 
only two photographs. 
  
Figure 6. A ray traced image of the Edicule (cupola and 
roof) CAD Model. 
5. STEREO-RESTITUTION, DIGITISATION AND 
GRAPHICS 
The exterior orientation elements of all photographs were 
loaded into the database of the Intergraph Intermap 
Analytic. Results of independent camera calibrations 
(focal lengths, principal point displacements and radial 
and tangential lens distortions) had also been stored at the 
IMA. Stereo models were set up using the IMA 
Simultaneous Orientation option, with principal point 
displacements and lens distortions (in look-up-tables) 
incorporated into the real-time loop. For Hassleblad 
photographs, 6 images were placed on each stage so as to 
allow 6 stereo models to be called up and interchanged. In 
this way several views of the complicated 3D structure 
could be used during the digitisation process. 
The initial digitising of features on the IMA was in the 
form of 3-D linestring elements, These were stored in 
individual design files, one for each stereo pair. These files 
were then merged, or referenced together, this allowed 
for representation within the CAD system of separate 
    
elevations, plans and cross-sections. All the data can be 
merged together, but with such objects as the Edicule, 
these unstructured, individual linestrings, whilst giving a 
good overall impression, cause confusion when studied in 
detail (Fig 5). They do however provide an accurate 
spatial representation of linear features which can be used 
to generate data which are easier to interpret and of more 
relevance to archaeological investigation. 
To make full use of 3-D CAD models of objects (Fig. 6), 
the individual features must be represented by shapes that 
have surface characteristics. These shapes have to be 
derived from the onginal linestrings (Littleworth et al 
1992). For ancient objects with their eroded and worn 
stonework, creating closed shapes from the original 
linestrings is a time-consuming practice. Assumptions 
must be made concerning the form of objects to enable the 
placement of CAD shapes and surfaces. This inevitably 
leads to lowering of the original photogrammetric 
accuracy, but the usefulness of the data is increased. The 
original linestrings (or the photographs themselves) can 
be referred to when accuracy is important, for example in 
checking alignments and height differences. 
6. CONCLUSIONS 
The secondary data from photogrammetry are 
conventional archaeological plans, elevations and 
sections and the three dimensional computer-graphics 
models of the Edicule. The latter are at present simplified 
but accurate representations of the main structural 
elements. The original decorative features are shown on 
the photography and can be plotted or digitised to give a 
full surface model if required. At present the main work 
is to combine the photogrammetrically derived computer 
graphics with the archaeological record in a database. The 
floor plans produced by archaeologists have been scanned 
and fitted to the photogrammetric data. Previous 
representations of structures on the site of and around the 
present Edicule are also being digitised and used in 
conjunction with modern data to see how the present 
structure might be related to earlier ones, and possibly to 
see to what extent remains of the original rock-cut tomb 
might be included in the walls of the present Edicule. 
Only by photogrammetry, a non-intrusive, passive 
technique, could such an immense amount of accurate 
detail have been obtained. Use of the data will lead to 
hypotheses about the present structure and its 
development from the original Constantinian Edicule. 
These hypotheses will it is hoped be of value to those who 
will be responsible for the renovation, restoration or repair 
of the Edicule. Such work can not be too long delayed, 
given the present parlous state of the building. 
7. REFERENCES 
Biddle, M., Cooper, M.A.R and Robson, S., 1992. The 
Tomb of Christ, Jerusalem: a photogrammetric survey. 
Photogrammetric Record 14(79) 25-43: 
Daniel, Abbot, 1106. The pilgrimage of the Russian Abbot 
Daniel in the Holy Land. Translated and annotated by 
Wilkinson, J., et al. In: Jerusalem Pilgrimage 1099 - 1185. 
Hakluyt Society (Second Series), London 1988. 167: 
pp187-207.