Íslenskar landbúnaðarrannsóknir - 01.03.1970, Blaðsíða 39

COLOUR INHERITANCE IN ICELANDIC SHEEP 37
When the rams had been genotyped in
this manner, a card was ntade ont for each
ram for each of his years on record. The
knowledge about his genotype for colour
in that particular year and his colour
score, if available, was punched into the
card, togetlier witli his identification. This
information was then transferred to the
proper fields in the data cards in a re-
producer, and when this was completed the
data were ready for final checking and
analysis.
D. TRANSFER OF CARDS
TO MAGNETIC TAPE
Soon after the card reader of the KDF9
computer of the Edinburgh Regional Com-
puting Centre (ERCC) had been installed,
preliminary programming for transferring
the card data on to magnetic tape was
started.
The ERCC had developed several rou-
tines in connection with reading cards and
packing contents of cards on to magnetic
tape as well as routines for recovering in-
formation from cards kept in packed form
on magnetic tape. These routines were
tried during the first stages of the pro-
gramming, but as tlie packing of the cards
by tlie ERCC routines was done in such
a way that the length of the record in
packed form varied, depending on the
contents of the card, it was decided not to
use the ERCC routines for packing.
Instead a special packing program was
written for this set of data, whereby each
card was packed into a record of constant
length.
The cards were read into store by using
the READ CARD BINARY routine, and
then 4 adjacent columns were packed into
one word, and each card thus packed into
20 words. Each 512 word block could
therefore hold 25 full cards, and the first
12 words of each block were not used for
packed information. The first word in
each block contained the block number,
the second the type of cards, the third the
number of the first card in the block and
the fourth word the number of the last
card in the block. The card numbers were
the consecutive numbers of the cards as
they were being read in. Words 5—12 in
each block were empty.
Sorne mechanical difficulties were en-
countered when the cards were being read
on to magnetic tape, and it was therefore
decided to develop routines which could
cope with all required types of analyses,
including sorting, using the cards in their
packed form on the magnetic tape so tliat
any further reading in of the cards could
be avoided.
When all the cards had been read on to
magnetic tape, they were scanned by a
special program which checked for double
punching in all numerical fields and
checked that fields with alphabetic in-
formation contained certain permitted
letters.
All cards witli faults found by the scann-
ing were printed out in full, and the cor-
rect values for the faulty columns found
if possible. In most of the cases the faults
were known to have been introduced by a
reproducer which hacl been used when
genotypes of sires were transferred to the
data cards. In cases of faults where the
correct values were not known with cer-
tainty, blank spaces were insertecl instead
of tlie faulty values.
When these corrections had been carried
out the data were ready for analysis.
E. SELECTION OF CARDS
OF INTEREST
Data where both alleles at the A-locus were
known in both sire and dam prior to a
particular mating were selected out of the