Water contents in healthy xylem, reaction zones (column boundary layers) and decayed wood were determined in sycamore (Acer pseudoplatanus L.) naturally infected by sooty bark disease (Cryptostroma corticale (Ellis & Everh.) Gregory & Waller) or wound-inoculated with Ustulina deusta (Fr.) Petrak. or Ganoderma adspersum (Schulz) Donk. Water contents determined gravimetrically were greater in reaction zones than in healthy wood by factors of 1·1–2·1 times. Maps of imageable proton density (M0), essentially equivalent to relative water contents, were also calculated from two independent nuclear magnetic resonance (NMR) imaging (MRI) data sets, acquired from T1 or T2-weighted experiments. M0 values derived from these, corrected for T2 and T1 respectively, were in good agreement, and paralleled the results obtained when total water was determined gravimetrically, although the proportional increase in M0 in reaction zones (1·3–3·7 times) was greater than the corresponding increase in total water.

Living wood saturated with water by vacuum infiltration contained approx. 1·3 times more water, determined gravimetrically, than uninfiltrated wood. The proportional increase in M0 was greater (1·8–2·4 times). This was attributable to the alleviation of magnetic susceptibility effects by elimination of the gas–water interfaces normally present in the uninfiltrated wood. The similarities between infiltrated wood and reaction zone tissues in both total gravimetric water content and NMR-visible water suggest that the biophysical environment of the water might be similar in both. This is compatible with the hypothesis that in reaction zones the gases normally present in the lumens of empty (dead) xylem fibres are replaced by an aqueous solution of phytoalexin-like compounds. This has been confirmed by microscopy. The implications of this in relation to antimicrobial defence and xylem function are discussed.

Although contrast in NMR images of lesions was predominantly attributable to changes in M0, a decrease in the spin–spin relaxation time (T2) was seen in the vicinity of some lesions. Spin–lattice relaxation times (T1) were not altered to the same extent.

Results obtained from both early and late stages of decay were much more variable than those from healthy wood or from reaction zones, and reflected both the low water content of decayed wood (and hence poor signal-to-noise ratios for NMR data) and the changing physical environment as the wood was degraded.