Sampling is often the most error-prone step in the entire analytical process. No matter how precise or sensitive the laboratory measurement, the result is meaningless if the sample does not truly represent the bulk material. Representative sampling ensures that every portion of the target population has an equal (or known) probability of being included, allowing valid statistical inference about the whole.
Several sampling strategies exist, each suited to different scenarios. Random sampling selects sampling points using a random number generator, eliminating selection bias. Stratified sampling divides the population into subgroups (strata) based on known characteristics — such as depth in a soil profile or location in a batch — and samples each stratum proportionally. Systematic sampling collects samples at regular intervals in space or time, which is efficient for process monitoring but risks alignment with periodic patterns. Grab sampling (judgmental sampling) relies on the operator’s experience to select locations and is the least statistically defensible approach; it should only be used when other methods are impractical.
Sample size has a direct impact on the uncertainty of the analytical result. The Ingamells sampling constant (K_s) relates the variance of the sampling process to the mass of the sample: w · R² = K_s, where w is the sample mass and R is the relative standard deviation. This relationship allows the analyst to calculate the minimum sample mass needed to achieve a target precision. Homogeneous liquids require far less sample mass than heterogeneous solids, where particle size and composition variability dominate the sampling error.
For solid sampling, techniques include cone-and-quartering, riffle splitting, and rotary splitting — all designed to reduce bulk material to a laboratory sample without bias. Particle size is reduced through crushing, grinding, and sieving. Liquid sampling requires consideration of stratification, mixing, and potential precipitation. Depth-integrated samplers or peristaltic pumps may be used for surface waters, while thief samplers are common for drums and tanks. Gas sampling involves collecting air or process gases into canisters, impingers, or adsorbent tubes, with careful attention to moisture, temperature, and reactive losses throughout.
Contamination prevention is paramount at every stage. Samples must be collected using clean, inert containers — glass, HDPE, PFA, or stainless steel depending on the analyte. Chain of custody documentation tracks the sample from collection through transportation, storage, preparation, and analysis. Proper labeling, sealing, and temperature control preserve sample integrity. A well-designed sampling plan, documented in a standard operating procedure, is the first and most critical link in the chain of analytical quality.
