We show that the two prism tunnelling experiment of Mizobuchi and Ohta
ke [Phys. Lett. A 168 (1992) 1] does not verify the quantum optical pr
ediction due to insufficient statistical precision in the anticoincide
nce measurement. We reanalyze their data and show that the observed nu
mber of coincidences is actually even larger than what is expected fro
m a classical coherent light source.