This paper examines the question of inference on three-dimensional lin
e process densities from intersection counts with cylinders, when isot
ropy of the line process cannot be assumed. It is shown how to modify
results derived for planar sections. A motivating application is the s
tudy of root development by inserting minirhizotrons into soil. Aspect
s of experimental design are suggested, such as placing tubes at an an
gle of 54-degrees to the vertical where they will be insensitive to th
e degree of anisotropy. The case where axial symmetry may not be assum
ed is introduced, and is shown to be considerably more complex. An ana
lysis of some experimental data is included.