It has been widely acknowledged that Doris Humphrey and Martha Graham were the two most influential exponents of American modern dance. Graham's work has been the more prominent, in part because she outlived Humphrey by thirty-two years and performed for a much longer period. This does not, however, diminish the influence that Humphrey's work has had on modern dance since her death in 1958. More significant, perhaps, is the influence it can have in the future. Humphrey's legacy includes a certain amount of documentary literature, including her seminal book on the craft of choreography, The Art of Making Dances (1959); photographs and film footage of her dancing and of her dances; and a codified dance technique which is taught on a wider scale now than ever before. The dances, however, need to be performed in the theatre; if they remain as an archive, they may be regarded as such, and the purpose of my work is to illuminate these dances for a contemporary audience. This article will discuss strategies undertaken to develop a perspective on modern dance production, including the significance of style; the search for a living past' drawing on the ideas of R. G. Collingwood; the identification, viewing and interpretation of evidence, including the use of a Labanotation score. The production processes employed by a range of artists involved in reconstruction will be considered, and my own practice positioned in relation to this. The notion of co-authorship will be examined within the contexts of these respective practices, illustrated by examples from recent Humphrey productions. Other performing arts have survived to a large extent through text-based evidence, but there is no immediate parallel existing in dance. A number of notation systems are utilised, including Benesh, Eshkol-Wachman and Labanotation, with the latter serving regularly for the recording of modern dance during the past fifty years. I would suggest that this, or any symbol-based system is not wholly comparable with those existing in music and drama, in part because the score is written by someone other than the choreographer. Despite the developments in Labananalysis, crucial aspects of movement quality and style, which are integral aspects of interpreting a work, are not in evidence within the Humphrey scores I have encountered. This is not a criticism of Labanotation, or other systems, rather a critical observation of notation.