Investigates the so-called stretcher effect, a phenomenon occurring in canvas paintings in which the canvas and paint layers in the regions closest to the stretcher, strainer or cross bars are in markedly better condition than those in other regions of the painting, behind which no wood is present to provide a buffer. The authors assert that this sharp transition results from local differences in relative humidity in the vicinity of the stretcher. Investigation focused on whether these relative humidity variations occurred hygroscopically or thermally, i.e. were induced by moisture or by heat emanating from walls. An experiment was devised that placed a mock-up painting against a gypsum wall in a walk-in climate room where relative humidity and temperature could be varied, and surface hygroscopic behaviour could be altered using plastic foil and cardboard sheets. Sensors measured air temperature and relative humidity at the wall, stretcher surfaces and canvas areas. Specifications on the test canvas painting and sensors are provided, and results presented in tables. Thermal gradients were found to cause significant variations in relative humidity near the canvas, and thus play a major role in the formation of the stretcher effect. Appendices provide models for hydroscopically and thermally induced RH differences.