The reliability of the automatic determination of the ABO group on the
Groupamatic is explained and backed up with an experience of 600,000 runs. Provided
the user observes the rules of conventional serology, this reliability depends upon three
essential, combined actions of the machine: identifying the sample, reading serological
reactions, processing and printing-out grouping data. Figures taken from 150,000 ABO
groups enable us to compare the frequency of detected weak A sub-groups with known
properties; tests performed on samples of chimeras or of Bombay phenotypes, associated
with experimental work on artificial chimeras, confirm the high reliability this automatic
system provides.