The geometric tortoise Psammobates geometricus, found in the Western Cape,
is one of the world's rarest tortoises. The cryptic colouration and sedenta
ry behaviour of the geometric tortoise causes problems when conducting coun
ts using standard capture-recapture methods. The primary purpose of this st
udy was to determine the detection probability of a geometric tortoise in t
he field on a purely experimental basis. Plaster of Paris tortoise models w
ere placed in predetermined transects in the natural habitat. Observers sur
veyed these transects for the tortoise models. Three group and two individu
al experiments were conducted. These experiments suggest that a large perce
ntage of geometric tortoises are not found during a census. In individual e
xperiments no observer detected more than 50% of the models, but in some ca
ses this percentage was slightly higher in group experiments. The effect of
the size of the tortoise, density of the habitat, level of experience and
searching ability of the observers are discussed. Estimates of the various
detection probabilities are derived and suggestions made for conducting tor
toise counts.