of a charged particle,
on many points along its trajectory; usually in order to determine its mass: an estimate of combined with momentum measurement, will allow one to put limits on the particle mass.
Ionization sampling is often combined with the measurement of the particle position, but may also be left to devices dedicated to the measurement of .
Typically, ionization sampling is done over small amounts of lost energy as in the gas of a drift chamber or in a bubble chamber liquid. Due to the fluctuations in the energy loss in thin slices, it is important to obtain a large number of samplings. It is then the statistical distribution of values measured for the same track which allows an estimation of the velocity 
, on which energy loss
depends. In analysing the sample of local values, care must be taken to use a sensible estimator: due to the tail in the Landau distribution,
a simple mean value will be a bad estimator. Mostly, one uses either a truncated mean (like eliminating the 20% of highest individual measurements), or
resorts to a full maximum likelihood treatment.
Ample discussion on optimizing detectors and readout for ionization sampling, and a vast amount of literature is also found in review papers, e.g. [Lehraus83], [Allison91]. [Blum93] discusses in detail the collection and analysis of ionization samples in drift chambers.