Mutations in the PERIANTHIA (PAN) gene of Arabidopsis thaliana specifically transform flowers from tetramerous to largely pentamerous, which is a characteristic of flowers of ancestral plants. We have cloned the PAN gene and here we show that it encodes a member of the basic region/leucine zipper class of transcription factors. Immunohistochemical analysis shows that the encoded protein is present in the apical meristem, the floral meristem, each whorl of organ primordia, and in ovule primordia during wild-type flower development. PAN expression occurs independently of genes affecting floral meristem identity, floral meristem size, or floral organ number. The near absence of a phenotype in transgenic plants overexpressing PAN and the contrast between the broad expression of PAN and the specificity of its mutant phenotype suggest that its activity may be regulated post-translationally or by the presence of partner proteins. Based on these results and on data reported previously, we propose models for the role of PAN in the evolution of flower pattern in the mustard family. Arabidopsis flowers arise initially as undifferentiated bulges on the flank of the apical meristem. Soon afterward, whorls of organ primordia arise sequentially. Four sepal primordia develop along the edges of the floral meristem, establishing the first whorl. Then four petal and six stamen primordia initiate in whorls 2 and 3, respectively, followed by development of two carpel primordia in the center of the floral meristem (Smyth et al. 1990).A proposed mechanism of floral organ primordium initiation and formation must account for how a primordium is formed and also when and where on the floral meristem it develops. The eventual fate of organ primordia is determined by the organ identity genes (Komaki et al. 1988;Bowman et al. 1989Bowman et al. , 1991Bowman et al. , 1993Hill and Lord 1989;Kunst et al. 1989;Irish and Sussex 1990;Jack et al. 1992;Weigel and Meyerowitz 1994); however, the position in which they arise (and thus their number) appears to be established independently (Meyerowitz 1997). How proper floral organ number is achieved is not well understood.One way to understand the molecular mechanisms by which appropriate floral organ number is determined is to find mutations that affect floral organ number but do not affect floral organ identity or floral meristem identity, and then to elucidate the nature and the function of the corresponding wild-type genes. There are a number of reports demonstrating the existence of such muta-, and wiggum (wig) (Leyser and Furner 1992;Clark et al. 1993Clark et al. , 1995Clark et al. , 1997Roe et al. 1993Roe et al. , 1997Sessions and Zambryski 1995;Talbert et al. 1995;Running and Meyerowitz 1996;Sessions et al. 1997;Laufs et al. 1998;Kayes and Clark 1998;Running et al. 1998). In clv1, clv3, and wig mutants, the increase in organ number is correlated with increased cell number in the floral meristem (Clark et al. 1993(Clark et al. , 1995Running et al. 1998). This indicates a mechanism by which the pos...