The purpose of this article is to clarify the relations between the way of spatial thought and the social structure in early medieval Japan by investigating the writing pattern of spatial representation of title deeds (shomon), including various types of documents such as baiken (land transaction document), shobun-jo (inheritance document), kishin-jo (donation document). The study area here is Yamato Province (present Nara Prefecture), where numerous documents have survived since the early medieval period. The study period (783 to 1250) are recognized as Heian and the first half of Kamakura Periods.For historians, such documents are the principal material for research. Recently, there has been a renewed interest in the attitude of people toward documentation of right in ancient and medieval Japan. Those studies tell us interesting facts, but almost all of them remain fragmented and discuss only some interesting aspects of documents. Except Kanno (1994), comprehensive research about the writing pattern of spatial representation of documents is lacking indeed.We need to refer to previous discussions concerned with land registers, basic documents of land management and those with land management in ancient and medieval Japan. In the discipline of historical geography, Kinda (1985, 1993) has achieved excellent systematic studies from this perspective: mainly based on empirical research of grid patterns (jori plan), he has elucidated the relationship with various types of small area place names, including the four-edges (shiishi) and territorial units such as county (gun) and "hundred" (go). It is considered that such a focus on the jori plan would be quite effective for clarifying institutional changes of land management.This category of research, however, has the drawback that different spatial representations are usually explained only with changing land management. Thus, different representations due to different actors as well as their regional differences tend to be difficult to examine. To further this standpoint, it is necessary to explore how the changing process of institutions was related to the changing process of document forms and in what representations the process was reflected.To deal with these problems, statistical approaches were selected. To begin with, matrices whose cells consists of binary data (0 or 1), need to be prepared from title deeds. The data sources here are Heian-ibun and Kamakura-ibun, both of which are compiled historical records. Generally, one sample was extracted from one document. In some cases, plural samples are extracted from one document, when the contracts of two or more places are recorded in one document. Through this extraction, 989 samples were prepared finally, and they were arranged in rows of the data matrices. In the columns, the attributes of the samples were arranged, and the attributes were classified into the two types: one concerns the spatial representations found in the documents like jori, shiishi, small place name and territorial units (gun, go and so on), and the other concerns the situation of document preparation such as period, area, social class of document maker, land size and land use.In the third section, the context of document preparation is discussed. It is quantitatively confirmed that the number of documents increased since the 12th century. This shows the penetration of the mentality that documentation of rights was essential in the alienation of property.In the fourth section, relations between each of the spatial representations and each of the contexts of document preparation are analyzed individually based on the two-dimensional contingency tables (see Table 2). As a result, Kindas schema that difference of period causes the variation of spatial representations, especially of jori, is confirmed quantitatively.