The paper outlines ways in which agricultural scientists fail to benefit fully from collaboration with statisticians and mathematicians and from the use of computers. These result mainly from misunderstandings and wrong attitudes; the aim here is to correct them by frank examination. A statistician can easily point comment on how his attempts to cooperate with others are rendered ineffective, and many examples are quoted. He must also recognize and seek to correct his own weaknesses; the paper has no intention of minimizing these, though its chief concern is with other matters. The emphasis throughout is on statisticians and mathematicians being collaborators in research, not merely purveyors of specialized services, on computer use being coherently planned, and on scientific management being concerned with priorities.