Tumour oxygenation is a critical determinant of success in radiotherapy and, indeed, it can be important in other forms of cancer therapy. Many attempts, extending over several decades, have been made to improve tumour oxygenation for therapeutic advantage, but no method has become established in routine clinical practice. A device for rapidly determining the local pO2 in tumours would clearly be very useful for research into these questions and perhaps also for assessment of radiobiological hypoxia in the clinic. A method that permitted continuous monitoring would be particularly valuable.

Over the past decade, oxygen electrode measurements of tumour pO2 have been performed routinely in many specialized centres using the Eppendorf pO2 electrode, a thin electrode (12 or 17 mm) contained within a 300 mm diameter bevelled steel needle [1]. This form of histography has a number of disadvantages. The principal disadvantage is the electrode consumes oxygen by electrochemical reduction, causing a continuous signal decrease with time. This can result in underestimation of the pO2 level. Measurements are therefore made while driving the probe through the tumour in a stepwise manner (termed pilgrim-like, as the electrode takes``two steps forward and one step back''), resulting in pO2 values from each stopping point in the electrode track through the tumour. A consequence of the need to keep advancing the electrode is that one cannot monitor timedependent changes in pO2 from a single location. Even if the electrode is removed and reinserted between readings, repetitive pO2 measurements from the same location may be erroneous, owing to bleeding, oedema and alteration of microcirculation in injured tissues. A more subtle problem is that the absolute reading depends on the O2 diffusion properties and oxygen solubility of the quite large volume (about 50ą500 cells, see below) surrounding the electrode tip. Tumours are usually heterogeneous, and only a few viable