The measurement of polarization of the fluorescence of solutions under high pressure is rendered uncertain by the photoelastic birefringency of the windows, and needs a correction dependent upon the fraction of the light emerging from a window with polarization normal to that of the incident light (scrambling coefficient). This correction, which increases rapidly with applied pressure, differs according to whether the method of measurement employs rotation of the excitation polarizer (T format) or the emission polarizer (L format), and the latter is shown to involve the more reliable and the smaller correction. A four window pressure bomb housing a 1 ml volume cuvette that permits absorption and fluorescence measurements up to 4 kbar is described. The polarization correction up to 2.5 kbar was determined. Its reliability was demonstrated by measurements of the fluorescence polarization spectra of fluorophores in solvents of high viscosity at all pressures and by comparison of isopiestic and isothermal measurements of a fluorophore in isobutanol, a fluid solvent with viscosity that changes rapidly with pressure or temperature.