The paper considers a question of how many sampling stations we should have to measure the phytoplankton biomass with given certainty due to the patch character of phytoplankton spatial distribution. It is shown that this relation between the station number and necessary certainty has a hyperbolic view and is resulted from routine formulae of the mathematical statistics. A compromise between the toil of the sample treatment and necessary precision of phytoplankton biomass measurement led to a following practice: a lake basin was covered by 15–17 stations that gave us a certainty of ±30 % while some measurement values could differ one from other on an order. The Birkhoff theorem is analysed on how it fits to Lake Baikal conditions and some recommendations are proposed in order to carry out measurements within the theorem validity.