106 EARLY CHOSON DYNASTYnot really a measure of land area but a constant measure of crop yield produced
by an area that varied from 2.25 to 9.0 acres, depending on the fertility of the
land. By the seventeenth century the average size of land parcels had decreased
so greatly that they were measured in pu, or hundredths of a kyo! because only
a minute fraction of landowners held more than one kyol's worth of land. Accord-
ing to a Japanese survey in 190 I, the average holding varied from 4.7 acres per
family for South ChOlla Province and 5.7 for North Kyongsang in the south where
the land was most fertile, to 6.9 for Kyonggi in the central region, and to 8.7 for
North P'yong'an, and 9.8 for South Hamgyong in the north where the land was
least fertile and dry farming was far more extensive than rice cultivation. 3 D
A land survey done in Kangwon Province in 1436 cited by Fukaya Toshitetsu
divided landowners by the size of their holding. The highest category, "grand
household" (taeho), included all those who owned over 50 kyol, and the aver-
age holding in that class was 80 kyo! (about 400 acres at about 5 acres/kyol). By
contrast, a smallholder (soho) was defined as one who held between 6 and 9
kyo!, and the average holding in this group was 7.5 kyat (37.5 acres). The low-
est category offragmentary smallholders held 2.5 kyo/ on the average (12.5 acres).
Kim Yongsop has pointed out that since Kangwon Province had less fertile land
than the south, the pattern of distribution in the south would have been more
greatly divided and the average holding smaller. He also cited other sources indi-
cating an average holding of one kyol (5 acres) per family for the fifteenth cen-
tury, but even in that case, it would have been twice as large as what Kim found
in his seventeenth-and eighteenth-century examples (2.5 acres).3'
Kim Yongsop conducted a study of 2,193 households listed in the yang 'an or
land registers of five districts in three provinces in the south in 1669, 1719, and
- The names of the heads of households in those registers were listed as
kiju, literally "the lord (or person) who cultivates," and Kim insisted that the
kiju had to indicate the owners of the land parcels listed in the yang 'an, even
though it was not the standard term for landowner. His main argument, that the
second character of the two-character compound, -ju, "lord" or "master," often
indicated ownership, as in the term, chiju (landlord, or landowner), was certainly
plausible but not totally convincing since it is also possible that the kiju was
simply the registered cultivator of the parcel, possibly with responsibility for
turning over tax payments to the district magistrate. If so, the kiju might or might
not have been the owner of the parcel, and the yang 'an records would not be
useful for determining the distribution of ownership at all.
Kim, however, insisted that his data revealed the growing fragmentation in
the pattern of landownership, but even if true, there is one anomaly with his find-
ings. It may well have been the fault of the restricted number of areas that were
available to him, but the largest holdings were relatively small, and there is no
sign of any of the larger holdings or estates reflected in many qualitative state-
ments in other sources. Kim explained that a single owner could have accumu-
lated a large estate by holding a number of small parcels in other villages or
districts that would not be listed in the records he studied, but one would have