There has been much, mostly inconclusive, discussion about how to define the household in a manner suitable for comparative purposes. Certain conventional criteria are not very useful in the Icelandic context, where it appears that a person could be attached to more than one household, where the laws suggest it was possible for more than one household to be resident in the same uncompartmentalised farmhouse; and where headship might often be shared. Definitions, for example, based on co residence or on commensalism do not jibe all that well with the pastoral transhumance practised by the Icelanders. Sheep were tended and milked in summer in high pasture at shieldings by servants and other household members who slept and ate there during the summer but who were still understood to be attached to the main lowland farm in which other household members lived the entire year. Still, both coresidence and commensalism are a big part of what an Icelandic household was about, but a certain definitional roughness and subjectivity is needed in order to accommodate native categories and conceptions. For the purpose of this study I consider a person's household to be where sjhe eats and sleeps most of the time and where, even when not sleeping or eating there, he or she is perceived to have some right or duty to do so. This kind of looseness will cause trouble in marginal cases, but it is fairly serviceable nevertheless. It also allows for the possibility of multiple-household membership; something the ethnographic evidence suggests should not be totally precluded by definitions all too often adopted, without much refinement from the census taker, and it takes better account of the demands of the native style of pastoralism. Although qualifications and modifications will emerge when we consider the sources, I adopt for convenience the terminology of household type settled on by the Cambridge Group. Households are either simple or complex. A simple household has as its base the conjugal family unit, that is, a married couple and their unmarried children, but it also includes households headed by a single parent with children as well as married couples without children. Complex households, on the other hand, are said to be extended if they include other relatives who do not form conjugal units of their own. They are joint or multiple1 if they are comprised of two or more related married couples, although to make sense in the Icelandic context, the class of married couples must include those living in 'loose marriages', i.e. open and regularised concubinage. Also, native classifications of multiple householding need not depend on the kinship connection between the married couples. I thus consider, contrary to the Cambridge typology, that a farmstead run as a unified economic enterprise can constitute a single household even if some of its members are not related or do not recite kinship as the reason they are housed together.
Miller, William I. "Some Aspects of Householding in the Medieval Icelandic Commonwealth." Continuity & Change 3 (1988): 321-55.