This type seems to have been produced exclusively in Egypt. For a long time its chronology and production centers remained very uncertain, and most authors would not assign the beginning of the production earlier than the third century A.D. Cahn-Klaiber, for her part, would date an early example from the late first century B.C. to the first century A.D. (Cahn-Klaiber 1977, lamp no. 123) and the earlier examples of her variants a and b to the second half of the first century A.D. (Cahn-Klaiber 1977, p. 164). Since excavations in the 1980s at the Roman fort at Mons Claudianus, in the Eastern Desert of Egypt, we know that the type was in production at the beginning of the second century A.D. It probably continued into the third and fourth centuries (Bailey BM III, n. to pp. 226–29; Bailey 1991; Knowles 2006).
The various existing main classifications (Petrie 1905, Bernhard 1955, Michelucci 1975, Cahn-Klaiber 1977, and Shier 1978) are worked out more on morphological and decor criteria than on the scarce archaeological data. Michelucci, for example, distinguishes three major variants: in the first and oldest, the cylindrical elongated bow-shaped nozzle shows the affiliation of the type with Hellenistic models; in the second, the lamp body has a pronounced oval shape, and the frog on the upper part is rendered in a realistic way; in the third, the oval body is more pointed near the nozzle, and the treatment of the frog is much stylized, sometimes disintegrated to a point of no longer being recognizable. Under the generic term of “frog lamps,” authors have also indiscriminately classified examples whose decors may have no connection with a frog, such as ovolos, petals, monkeys, dogs, human embryos, heads of humans or of divinities (e.g., Bes), theater masks, ears of grain, and other images.
The bases of frog lamps are generally flat and may bear various potter’s marks. Among the most common: irregularly placed incuse notches, sometimes in the shape of the letter alpha.
The period of production of frog lamps appears to be long. Bailey mentions later variants of the fifth, sixth, and seventh centuries A.D. (Bailey BM III, pp. 227–28).