In Fechner's psychophysics, the 'mental' and the 'physical' were conceived as two phenomenal domains, connected by functional relations, not as two ontologically different realms. We follow the path from Fechner's foundational ideas and Mach's radical programme of a unitary science to later approaches to primary, psychophysically neutral experience (phenomenology, protophysics). We propose an 'integral psychophysics' as a mathematical study of law-like, invariant structures of primary experience. This approach is illustrated by a reinterpretation of psychophysical experiments in terms of perceptual situations (...) involving a constructed apparatus and an instructed subject. The problematic notion of 'measurement of sensation' is thus eliminated: 'sensations' are merely indices for classes of perceptually equivalent configurations (states of the apparatus) specified by the instruction. The locus of the measured is in the inter-subjectively shared, communicable world—not inside the subject's mind. Finally we discuss the role of integral psychophysics as a scientia prima , logically and methodically preceding physics and psychology. (shrink)