The task to be carried out by spectrometer is to determine the chemical compositions of samples. Anything done wrong “ahead of the spectrometer” during sample-taking and sample preparation cannot be rectified “behind the spectrometer” by computer, no matter how fast it may be.  Here I would like to appeal to all spectrometer users to devote more attention to this area “ahead of the spectrometer” which, with increasing data processing, with networking, statistics, works certificate etc. has been pushed into the background although it is precisely in this area “ ahead of the spectrometer” that even greater progress can be achieved, namely automatic sampling, sample preparation and automatic machine manipulation for automation sample transport to the spectrometer. This is where a good deal of manual and in some cases monotonous work is still being done, and because of its monotonous the necessary care quickly changes to neglect, eg:

-         the molten metal sample is deslagged,

-         the sample is not sufficiently deoxidized,

-         the sample is cast into an oxidized mould,

-        the sample is ground “blue” ect. ,

-        dissolution is not out at a sufficiently high temperature,

-         the solution is not filtered,

-         the solution is not used while new etc.,

-        contaminations may take place.

An automatic system works through the various steps uniformity and reproducibly, which can be proved from the uniformity of analyses over a lengthy period.

Selection of the correct samples for calibration, recalibration and control, also of material to be analysed and its preparation are part of the analytical procedure. A correct description and adherence to the foregoing are essential precondition for obtain precise and correct analyses.