ST. THOMAS AND ST. JOHN


XII, b). There were no differences in griddle forms with depth.
One, from Level 2 of Test III, did not have a raised edge.
 In analyzing the pottery, it was noticed that the grit of Botany
 Plain sherds was finer grained than that in similar sherds from
 Magens Bay. Surfaces, while irregular were extremely well burn-
 ished and "slick feeling." Burnishing was much more common and
 better done at Francis Bay than at any of the other sites.
 Presence of a fairly large amount of Coral and of Harbor Plain
 strongly suggest habitation at Francis Bay during the Magens I
 period. This appears to be the case in spite of the fact that more
 of these sherds were at shallow than at greater depths in the mid-
 den. Otherwise, the pottery suggests occupation during Magens
 IIB times. Probably the site was first inhabited in the Magens I
 period and occupation continued into the lagcnis IIB period.

 Tests at Coral Bay

 The Coral Bay site (Fig. 2), while probably smaller in area
 than the one at Magens Bay, is the largest one the island of St.
 John. Hatt worked extensively here and this site with that at Cruz
 Bay (to be mentioned later) and the Longford and Sprat Hall sites
 on St. Croix form the basis for Hatt's (1924: 31) earlier period.
 The catalog at the Danish National Museum indicates the midden
 was fairly shallow as only two levels are recorded. Our tests (Table
 5) also found the cultural deposit to be fairly shallow.
 The site has been cultivated for a great many years. At the time
of our investigation, much of it was in pasture. The soil is mostly
clay and ground cover, due to lack of moisture, extremely sparse.
Size of the Coral Bay site is hard to dtitenmine but five acres might
be a fair estimate. This site and the one at Cruz Bay are the only
sites of any size on either St. John or St. Thomas which are located
any substantial distance from salt water.
 Our investigation consisted of surface collecting, the digging of
many small holes in an endeavor to find a deposit suitable for
stratigraphic testing, and the excavation of four small tests. Feeling
our sample from Tests I and II was too small for this important
site, it was revisited and tests A and B were made. Results in
terms of pottery types are given in Table 5 and illustrated in