The abundance and recruitment of oak in northeastern US forests have long been topics of speculation. While physiology and life history suggest that oak recruitment benefits from disturbance, records of disturbance that predate European influences are largely absent. Tree ring records in upland stands do not extend beyond the early 17th century, when disease and warfare depopulated much of the eastern forests (White 1991; Clark 1995). Traditional sediment analyses rarely permit reliable identification of individual burns (Patterson et al. 1987; Clark 1988a; Clark & Patterson 1997). The paucity of evidence admits lively debate on past fire importance (Hawes 1923; Raup 1937; Day 1953; Russell 1983; Myers & Peroni 1983; Patterson & Sassaman 1988). Thin section methods have recently been developed that allow reconstruction of annual fire records (Clark 1988b, 1990) and this motivated us to reconsider pre-European fire at Devil's Bathtub, NY. These methods produce consistent estimates (i) in different cores from the same lake (Clark et al. 1989), (ii) among lakes within the same landscape which can then be matched with independent fire records (Clark 1990), and (iii) within and among regions, where variations are consistent with differences in climate and vegetation (fuel load) (Clark & Royall 1995a; Clark et al. 1996a,c). Archaeological evidence agrees with increased sediment charcoal during the rise and decline of Iroquois agriculture (Clark & Royall 1995b). During the period when spruce-pine dominated at Devil's Bathtub, charcoal peaks, which occur at 80-200-year intervals, are of the same magnitude as those observed at other sites during fire years. Charcoal peaks cease abruptly with early Holocene spruce-pine declines and the expansion of hardwoods at Devil's Bathtub; no peaks occur thereafter. This lack of any fire over millennia ran counter to our expectations, and we discussed caveats and the need for more studies (Clark & Royall 1996; Clark et al. 1996b). Abrams & Seischab (1997) feel we challenged 'the fire and oak hypothesis' (Abrams 1992) and list (i) how we overlooked palaeoevidence showing fire to be important in oak forests, (ii) reasons why our lake gives biased charcoal records, (iii) reasons why the site must be unrepresentative of the kinds of oak forests Abrams (1992) meant, and (iv) why the methods are flawed and how we misinterpreted our data. I address each of these four points in turn followed by a different perspective on data evaluation.