Stagnant nutrient solution containing 0.1% agar and with an extremely low oxygen level ('stagnant agar solution') was used to simulate the gaseous composition and slow gas diffusion of waterlogged soils. Comparisons were made between the growth of two wheat cultivars (Triticum aestivum, cvs. Gamenya and Kite) and one triticale cultivar (Triticosecale, cv. Muir) grown in stagnant relative to aerated solution. For all genotypes tested, immersion of roots in stagnant agar solution resulted in the death of the entire seminal root system and led to profuse branching of the laterals of the nodal roots. In the stagnant agar solution aerenchyma, as a percentage of the total cross sectional area of nodal roots, was 18% for Muir, 14% for Kite and 12% for Gamenya; the roots of species with more aerenchyma also attained a longer maximum root length as predicted by the model of Armstrong (in: Woolhouse HW, ed. Advances in botanical research, vol. 7. London: Academic Press, 1979). Muir also had a nodal root/shoot fresh weight ratio of 0.5 compared with 0.2-0.3 in Kite and Gamenya. The greater number and length of nodal roots of Muir did not lead to better shoot growth than in the other genotypes; one possible reason for this lack of improvement is a low efficiency of aerenchymatous roots in wheat. (C) 1998 Annals of Botany Company.