RESUMO
The aim of this study was to evaluate the effect of different sources of variation in gas production technique on the in vitro gas production kinetics of feedstuffs. Triplicates of commercial concentrate, grass silage, grass hay and grass pasture were incubated in three experiments: experiment 1 assessed two agitation methods; experiment 2 evaluated different rumen inocula (pooled or different donor cows for each incubation run); and experiment 3 used Goering-Van Soest or Mould buffers for media preparation. Gas production data were fitted into the Michaelis-Menten model and then subjected to analysis of variance. Gas production (GP) at 48 h and asymptote gas production (A) were lower when bottles were continuously under horizontal movement. Time to produce half and 75% of A, and A were affected by rumen inocula, while buffer type affected time to produce half and 25% of A and GP. No interactions between substrates and sources of variation were observed, suggesting that the effects of substrates on GP parameters were not modified. It is concluded that comparison of numerical data from in vitro experiments that follow different protocols must be done carefully. However, the ranking of different substrates is more robust and less affected by the sources of variation.