Application of the aperture mass ( M _ { ap } - ) statistics provides a weak lensing method for the detection of cluster-sized dark matter halos .
We present a new aperture filter function and maximise the effectiveness of the M _ { ap } -statistics to detect cluster-sized halos using analytical models .
We then use weak lensing mock catalogues generated from ray-tracing through N -body simulations , to analyse the effect of image treatment on the expected number density of halos .
Using the M _ { ap } -statistics , the aperture radius is typically several arcminutes , hence the aperture often lies partly outside a data field , consequently the signal-to-noise ratio of a halo detection decreases .
We study these border effects analytically and by using mock catalogues .
We find that the expected number density of halos decreases by a factor of two if the size of a field is comparable to the diameter of the aperture used .
We finally report on the results of a weak lensing cluster search applying the M _ { ap } -statistics to 50 randomly selected fields which were observed with FORS1 at the VLT .
Altogether the 50 VLT fields cover an area of 0.64 square degrees .
The I -band images were taken under excellent seeing conditions ( average seeing \approx 0 \hbox { $ . ^ { \prime \prime } $ } 6 ) which results in a high number density of galaxies used for the weak lensing analysis ( n \approx 26 { arcmin } ^ { -2 } ) .
In five of the VLT fields , we detect a significant M _ { ap } -signal which coincides with an overdensity of the light distribution .
These detections are thus excellent candidates for shear-selected clusters .