We present an analysis of the proximity effect in a sample of ten 2 Å resolution QSO spectra of the Ly \alpha forest at \langle z \rangle = 2.9 .
Rather than investigating variations in the number density of individual absorption lines we employ a novel technique that is based on the statistics of the transmitted flux itself .
We confirm the existence of the proximity effect at the > 99 per cent confidence level .
We derive a value for the mean intensity of the extragalactic background radiation at the Lyman limit of J = 3.5 ^ { +3.5 } _ { -1.3 } \times 10 ^ { -22 } ergs s ^ { -1 } cm ^ { -2 } Hz ^ { -1 } sr ^ { -1 } .
This value assumes that QSO redshifts measured from high ionization lines differ from the true systemic redshifts by \Delta v \approx 800 km s ^ { -1 } .
We find evidence at a level of 2.6 \sigma that the significance of the proximity effect is correlated with QSO Lyman limit luminosity .
Allowing for known QSO variability the significance of the correlation reduces to 2.1 \sigma .

The QSOs form a close group on the sky and the sample is thus well suited for an investigation of the foreground proximity effect , where the Ly \alpha forest of a background QSO is influenced by the UV radiation from a nearby foreground QSO .
From the complete sample we find no evidence for the existence of this effect , implying either that J > 20 \times 10 ^ { -22 } ergs s ^ { -1 } cm ^ { -2 } Hz ^ { -1 } sr ^ { -1 } or that QSOs emit at least a factor of 1.4 less ionizing radiation in the plane of the sky than along the line of sight to Earth .
We do however find one counter-example .
Our sample includes the fortunate constellation of a foreground QSO surrounded by four nearby background QSOs .
These four spectra all show underdense absorption within \pm 3000 km s ^ { -1 } of the redshift of the foreground QSO .