Conductivity, temperature and pressure profiling instruments usually being fitted with the best of available sensors, differences of readings with laboratory tests or intercalibration in sufficiently homogeneous water are merely a question of calibration, not of quality. Even in areas of very homogeneous deep water one finds random deviations of water properties much larger than the tolerance of measuring and calculation, and larger with the primary readings

,

and

than with the secondary ones

and S. These deviations give rise to additional large errors in the secondary readings, caused by the different transient responses of the sensors, and by the way to get the primary readings from the output of the sensors. By computer simulation it was found, that

C random deviation and a certain thermometer cause

ruggedness of

with a sampling system. With an averaging system the value is

and can easily be reduced to

.