Page 10 - 6686

average proportional value between the values of measure in the
first and second comparison, does not depend on the transmission
coefficient of measuring scheme. So constant systematic error of
this coefficient, which happens to be measured in a single
measure, is fully removed.
Random error
Random errors appear after random and unexpected changes
of the properties of tools and measurement conditions and
properties of sensory organs of the observer. They can be caused
by imperfection of method of measurement, the lack of
substantiation of theory or admitted simplifications, as a result not
only values, but also signs of errors remain unknown. Random are
errors undefined by its value or insufficiently studied errors, and it
is not possible to establish regularity of their appearance. They are
determined by a difficult aggregate of reasons, that are difficult to
analyze. Their values can not be predicted, and it can be
established regularity for frequency of appearance of their different
values only. Presence of random errors (unlike systemic errors) is
easily detected by repeating of measurements. Mostly, appearance
of random errors is a stationary random process.
If the value of random variable can form discrete (finite or
infinite) series of numbers, then this random variable is called
discrete. If the value of a random variable fill the whole interval
(finite or infinite), then the random variable is called continuous.
Each value of the random variable х r of discrete type
corresponds to a certain probability of its occurrence р. Each
interval (a, b) from the range of values of random variable of
continuous type also corresponds to a certain probability p {a <x
<b} that the value of a random variable will be in a certain
interval.
Ratio that establish connection between the possible values
of random variables and their probabilities, called the law of
10