Considering the general, though limited distribution of sites, it is remarkable that so few exhibited any considerable depth of deposit. There are localities in the southeastern United States where middens of great depth, over twenty feet, are known. Hence, it is surprising that middens of more than one foot in depth are quite rare in northeastern North Carolina.

There were a few places where it would have been quite revealing to have had middens of some depth. One of these, of course, would be Roanoke Island. Harrington (loc. cit., p. 145) had already had this frustrating experience. In this study nothing new was found to develop (Page 96) further the sequence of Roanoke Island aboriginal events. Hatteras Island was another such place, but here a little more promise was apparent. At the Cape Creek Site, H 1, the midden was less than two feet deep but its richness gave some promise of reward. Hence, a test area ten feet long and five wide was excavated in the most propitious spot, that is, the least disturbed. There has been no construction, plowings, or previous archeological investigations to disturb the site but much has been removed for road metal and two big cuts have been made through the site by roads. There is no way of estimating the extent of damage done by the winter storms that attack the Sound side of the island.

The results of the test excavation at H 1 are summarized in Table 3.

(Page 97) Not much may be learned from this excavation. It is a homogeneous site with shell-tempered pottery. Ninety-four per cent of the sherds are simple stamped. This clearly places the site in a late content. Simple-stamped (Page 98) ware has been "strongly suspected" of belonging within the historic period (Harrington, 1948, p. 252). On the other hand, not much stratigraphy could be ascertained at H 1. Apparently the few sand-grit-tempered sherds were not earlier than the shell tempered.

Of course, a more extensive excavation may have been more revealing. Certainly the presence of the pipe stems and a gun flint further suggest early historic contact. It would have been most desirable to have found evidence here of English objects, for this is the logical place to find remains of the Roanoke colonists. Quinn (1955, p. 908) believes this is the locality at which was found a "counter" similar to those excavated at Fort Raleigh (Porter, loc. cit., p. 35). However, as described under site H 7, it is at this latter site where the counter seems to have been found.

A second stratigraphic test was made at the Whalen site, P 1. Storms in 1954 had cut away about twenty five feet of the site and serious damage was done again in 1955. The site has been known for some years as an Indian village, and superficial examinations indicated a sample here might be illuminating.

A square five by five feet was selected in a part of the site which was thought never to have been plowed. The top three inches was mixed, of course, but the next six inches was nearly solid shell. Beneath that the shell was nearly absent although potsherds were more numerous. The soil gradually decreased in humus content and was quite clean at a depth of two feet.

The distribution of the pottery recovered in this test is tabulated below.

There is no doubt about the cultural relations of site P 1. It is typical of the many shell heaps as may be observed from Tables 1 and 2. Shell-tempered pottery may be present but it is always a minority ware. Further, the shell-tempering technique appears to be a late arrival, or at least more recent, an these sites.

The site that gave greatest evidence of depth was the Bandon Site, A 12, on Chowan River. Also this site was farthest northwest of all examined. Thus it might show contacts with cultures identified in the interior.

A pit five by five feet was excavated by 3-inch levels and, as in all other stratigraphic tests, the contents of each were screened from top to bottom.

(Page 100) An idealized profile of the Bandon Site shows that shell is confined to the top one foot or so without any great concentration. However, this is not typical of the site. Our test pit was purposely put down in an area where shells were less frequent. The soil was mixed humus, sand, ash, and charcoal fragments to a depth of four feet. Below that a pit penetrated the clean yellow sand for an additional one foot. The material recovered from this pit is listed under "pit" in Table 5.

(Page 102) A check of Table 1 shows that of the A 12 surface sherds none was shell tempered nor were any simple-stamped sherds found. These two facts demonstrate the cultural position of the site to be quite different from H 1 and other Banks. There is a definite decline in the use of net-impressed pottery that suggests the site to have a history that spans several centuries but persisted until near the eighteenth century.

Nothing else of significance came from the test. Pig bones were found in the top nine inches as well as were bricks, plaster fragments, glass, and crockery. This modern material is derived from some former nearby slave quarters. Numerous pieces of trade pipes came from the same top zone.

A second area of interest is that just north of Duck. Because of possible associations with the colony of Roanoke Island, the region was carefully examined. Two test pits were dug at B 5. Not much midden is apparent at the site but two pits five by five feet were sifted to a depth of two feet where sterile sand was encountered. In test area 1 the pottery fragments totaled 28, all of shell-tempered, fabric-impressed type. In test area 2 the sherds were shell tempered also; 18 were fabric impressed and 25 were plain, although the latter were so badly etched from wind erosion as to preclude any certainty about surface finish. Some, if not all, of these sherds may have been fabric impressed. The tempering material was completely leached from all specimens in both tests. Many specimens of chinaware found in the immediate area make the site of interest in other connections (see Figures 15 and 16)

Again, all that was ascertainable from these tests was that B 5 was a homogeneous site. The only surprising element is that no simple-stamped type was found.

Site A 1, at the tip of Currituck Peninsula seemed a likely spot for a stratigraphic test since surface finds showed a good representation (Page 104) of the shell-tempered ware and of grit-tenipered also. However, a careful test revealed that the cultural remains were concentrated in the top few inches and no stratigraphic separation could be developed. A 1 was once a very large and probably deep site, but the bulk of it has been removed to construct the fill for the highway leading to the bridge over Currituck Sound.

From the above excavations certain generalizations can be made. These are not very profound at first glance, but archoologically and historically some significance is involved. First, it is apparent that sites on the Outer Banks are of different content than those on the mainland. Duck, B 5, and Cape Creek, H 1, are judged to be contemporaries on the basis of shell-tempered ware being nearly the only ware represented at them. Site A 12 and Site P 1 are certainly completely independent of the cultural influences that affected the Outer Banks sites. The definitive answer to the question of sequence of pottery types was lacking, but the few straws reveal the direction of the wind--namely, that the shell-tempered ware is latest and thus sites on the Banks proper are among the most recent.