(…from Bulletin of the Society for Pennsylvania Archaeology, Inc.  Publication date ca. 1963)
 

THE MIANUS GORGE ROCK SHELTER†

By Bernard W. Powell*
 

Abstract

Salvage excavation of a rock shelter in the Mianus River Gorge of southwestern Connecticut revealed an occupation extending, without discontinuities, from a pre-ceramic to a ceramic horizon.  Although the recovered artifacts are few, the projectile point forms seem to parallel the transition from narrow stemmed forms to broader side-and-corner-notched forms, reported for the Hudson Valley some 30 miles to the west.  The site provides the first evidence of such an Archaic development in southwestern Connecticut.  A few potsherds in the upper level of the site have similarities to both the Windsor and East River wares, and may represent an early stage in the ceramic period.  Shell tempering is absent; a few sherds may be fiber tempered.
 

FROM ITS SOURCE in Westchester County, New York, the Mianus River flows into Fairfield County, Connecticut, and empties into Long Island Sound at Cos Cob Harbor, upon whose western shore is the Indian Field Site, described previously by the author (Powell, 1958).

     Along its middle and upper reaches, the Mianus flows through a ravine known locally as the Mianus Gorge (see Figure 1). Dominant rocks are quartzite and schist of the Hartland Formation (Rodgers, Gates, et al., 1956).  The geologist Flint (1930; p. 146) characterizes the region thus: "…extending up . . . Mianus River for considerable distances, is a group of terraces at 210 feet, characteristic in form and fringing ill-drained meadows."  His map shows these terraces formed in a local ice lake during Pleistocene times.

Fig. 1

  Notable botanicals are stands of Eastern or Canadian Hemlock (Tsuga canadensis), obviously of respectable antiquity.  The area has attracted the attention of naturalists, one of whom is quoted: "... a place in which will be found the unaltered flora and fauna of the Northeast" (Stern, 1958).  It seems that the region must resemble to some extent the virgin woodlands of pre-Contact Connecticut. 1

     Around 1950, I began intermittent archeological reconnaissance in this watershed.  I was stimulated to this partly by the pioneering work of Harrington, who a half century ago found and described several rock shelters not far west of here, near Armonk, N. Y. (Harrington, 1909).  Ultimately, I discovered at least one rock shelter suggesting multiple occupancy, and formed a general impression of the entire area; this report is the result. 2

     Part of the land surveyed by me (including the site of the rock shelter) belongs to a local water company, which in 1954-55 constructed an earth dam across the Mianus, subsequently inundating a large portion of the Gorge.  Almost certainly, sites of Indian occupancy along the river bottom were thus destroyed.  The loss of possible stratigraphic deposits is the familiar "archeological calamity" of one archeologist appalled at commercial practices and destruction in a Connecticut rock shelter (MacCurdy, 1914; p. 522).  Prior to flooding, at least one locality - the only really open field along the original bottom of the Gorge, right astraddle the state line - yielded tiny grains of charcoal at depths to 4 feet in test holes, and one tiny fragment of marine clamshell (probably Venus mercenaria).  The spot is now under 15-25feet of water, but may mark a cornfield or small village site of Woodland times.

     Hikers and others, over the years, reported finds of "arrowheads" and places with "lots of clam shells" throughout this (NB: words missing, apparently also in original) unproductive. The Gorge is rocky, wild, and heavily overgrown.  Poison ivy (Rhus) abounds, and a copperhead (Agkistrodon contortrix) infestation is well established.

 The first hint of a substantial relic-bearing deposit came from the rock shelter, where a few quartz chips and some fragments of the fresh-water mussel (Unio) were noted weathering from the soil on the shelter floor.3  This shelter (Figure 2) is formed by an overhanging rock ledge, and faces almost due east, part way up the side of the Gorge.  The site is three-quarters of a mile northeast of the southwest corner of the Pound Ridge Quadrangle map (U.  S. Geological Survey, 1951). It is about 260 feet above sea level; originally it was about 50 feet higher than the level of the stream, but the impounded waters of the reservoir now lap at the very base of the slope beneath the shelter.

 

 

EXCAVATION

 

     This site, which I call the Mianus Gorge Rock Shelter, merited careful excavation, so a standard 5-foot grid oriented on magnetic north was laid out embracing the major portion of the shelter (see Figure 3).   v:shapes="_x0000_s1026"> Several squares were excavated in whole or part to a depth of 18 inches, and one test pit (C in Square ON/IOW) was carried below this to 36 inches.  Certain problems relating to deposits at depth in Finch's Rock House, the largest rock shelter excavated by Harrington (op. cit.) prompted this sounding.  No artifacts were recovered from this test pit, nor was there evidence of a "darker" layer at depth, as reported at Finch's Rock House.  Barely discernible charcoal grains and tiny quartz flecks were noted with decreasing frequency to the lower part of this sounding.  It was not feasible to dig deeper at the time.  Anticipated permission to extend my investigations was not forthcoming from the landowner, so my work was limited to a few days. 

 

Since the materials recovered were, however, removed under control, they merit description and possibly permit a few tentative conclusions.4

      Strong soil profiles were not encountered.  Figure 4 may suggest sharper breaks than actually occurred.  The usual surface litter of leaves and leaf mold graded imperceptibly into rich, black topsoil about 6 inches thick.  This in turn gave way to tan, silty subsoil which became granular with depth.  Large rock slabs (the subsurface rock indicated in Figure 3) underlay much of the floor and prevented further excavation.  Perhaps these slabs represent ancient rock-falls within or upon the shelter.  Their size precluded complete investigation beneath them, but in most cases the soil appeared sterile.

v:shapes="_x0000_s1027">    Without clear physical stratigraphy, then, specimens were arbitrarily assigned to 6-inch levels, of which I recognize three in my 18-inch depth: 0-6 inches, hereafter called Level 1; 6-12 inches, hereafter called Level 2: and 12-18 inches, hereafter called Level 3. For reasons which emerge shortly, Levels 2 and 3 are treated together in the Discussion section of this paper.

     An almost complete lack of bone or shell refuse in any of the layers dug in the shelter floor prompted me to dig a trench (Trench A in Figures 3 & 4) in the slope at the front of the shelter.  This seemed logical as a place for refuse to be deposited by the former inhabitants, but the trench, dug to a depth of about 18 inches for over 8 feet down the slope, yielded only one small rectanguloid quartz scraper (specimen gg in Figure 5). This was recovered at a depth of 8 inches at the upper end of the slope.  Rocks encountered in the floor of the trench precluded further digging.  It is interesting that Harrington (op. cit.) in writing of his work at Finch's Rock House mentioned a ". . . fan-shaped dump of earth at the north entrance, the  'sweepings' of the ancient dwellers," and further that this dump was stony and only 18 inches deep.

 

v:shapes="_x0000_s1028">Two small fire pits were noted; these were Fire pit A in Square 5N/IOW and Fire pit B in Square ON/5W, both indicated in Figures 3 & 4. Both were irregular, bowl-shaped features no more than 15 inches across, and both extended from Level 1 into the layers below, B being about 15 inches deep.  Ashy-textured, fluffy soil charged with charcoal grains filled these fire pits.  They were not rock-lined.  A stemmed projectile point of fine gray siliceous stone was associated with the lowest level of Fire pit B at a depth of 15" (specimen p in Figure 6).

     Behind the rock mass forming this shelter is a large crack paralleling the long dimension of my excavation.  This cranny seemed a logical place for a cache, or even a burial.  However, investigation disclosed neither artifacts nor organic remains, and I conclude the slot was not used by the Indians.

 

     The high point of the shelter floor is at the south end, and in profile (see Figure 4) the floor drops about 2 feet to the northern end.  Rain-wash has apparently always been impounded south of the 5N line, since the shelter narrows here, and even deposition of in-washing material has been prevented.


     The datum point was permanently fixed by an "X" chiseled in the rock.  Further description by individual layer follows.

Level 1

     As noted, the shelter was discovered when a few quartz chips and a tiny, eroded potsherd were discovered beneath the maximum overhang line (Figure 3), where rain and ground water drip from above had exposed them.  The only evidence suggesting white men ever visited the shelter was a .32-caliber brass cartridge case found among the leaves on the shelter floor.  After these leaves were removed, excavation proceeded uniformly to a depth of 6 inches.  Most material from Level I came within the first few minutes of excavation in Square 5N/ 10W.  This most inward recess of the shelter was logically an optimum place for deposit of cultural refuse, saving the slope in front which, as mentioned, proved largely negative to my small exploratory trench.

     The total projectile point inventory from Level 1 is given in Table 1. The chert point (f in Figure 6), was finished with fine, controlled pressure flaking.  All points are bifacially flaked; some have sinuous edges where flakes were removed alternately from each side.  The stemmed points (m and r in Figure 6) show some asymmetry.  Asymmetry in side-notched and stemmed points was noted by Patterson at the NAS Site I, Garvie Point, Long Island, N. Y. (Patterson, 1955; p. 2).  This site is 20 miles south of the Mianus Gorge Rock Shelter.  Byers says asymmetry as a feature ". . ..is not infrequently found among points in southern New England" (Byers, 1958; p. 2).  These same two projectile points may show incipient basal grinding. v:shapes="_x0000_i1028">

Over 200 miscellaneous cores, fragments, and flakes were recovered from Level 1. A rough breakdown of these is shown in Table II.  Most of this material undoubtedly is workshop trash.  However, many of the flakes have sharp edges and could have been utilized as scrapers.  I believe that oftentimes this was done, though examination of the collection reveals few patinated, worn, or "flaked back" pieces.    

Researchers at a Midwestern bison-kill site observed that expediency might dictate use and manufacture of crude scrapers in a hunting site even by those with the ability and the cultural framework to create refined forms (Agogino & Frankforter, 1960; p. 415).  Perhaps 20 of this lot come nearer to being scrapers than the others; at least one (gg in Figure 5) is a recognizable rectanguloid plano-convex scraper.

v:shapes="_x0000_s1029">    The presence of both fresh-water and marine shellfish valves in this layer (see Table III) was interesting, though the amount recovered was extremely small.  The shelter is 8 miles from the nearest salt water.  None of the potsherds were shell-tempered, and no artifacts of shell were noted.  That the shell therefore represents food scrap seems a justifiable conclusion.

     Charred nutmeats were also recovered but they may have been introduced to the shelter by rodents, and charred in situ by natural brush fires.  At least this possibility cannot be precluded.

 

     Only 150 rather badly degraded potsherds were recovered.  All save one came from Level 1. This deviant, recorded from Level 2, is, I believe, either intrusive into it, or else improperly recorded.  The state of the ceramic inventory precluded any really meaningful analysis, but such data as were derived are given in Table IV.

Levels 2 and 3

    Since the supposition is entertained in the Discussion section of this paper that Levels 2 and 3 may be combined, their projectile point inventories are so treated in Table V. All points are bifacially flaked; they are also illustrated in Figure 6. About half the points are white quartz.  Removal of small flakes by skilled pressure retouching is evident along the edges of j. The chert point e has a bi-furcated base.  The tip of the white quartz point q is broken.

     The specimen labeled w is an interesting one.  It is made of a brown-gray silicate with a fracture partly conchoidal, partly "slatey" (hinge-like).  This latter type of fracture may be controlled by obscure planes of bedding or cleavage, persisting even after metamorphosis of the stone.  A very similar condition was cited by Byers for the quartzite called "Moosup Stone," present at the E. D. Prey Site in eastern Connecticut (Byers, 1958; p. 2).  Very fine pressure flaking along the edges of this specimen gives a serrated appearance, as seen in the illustration.

The stemmed point, p, was cited previously as present in the v:shapes="_x0000_s1030">lower level of Fire pit B. Although it is listed here, it may be intrusive, since Fire pit B extended down from Level 1.

Two very large specimens, not included in Table V, are x and y in Figure 6. These were probably spear or lance points, but they may have been hafted knives, so are not included as projectile points.  Specimen x is a dark greenish flint whose tip was broken during aboriginal times.  It was later reworked along the break, as revealed by fine pressure-chipping marks.  The specimen was then no doubt used as a knife or possibly a hafted scraper.  It might be classified as corner-removed in manufacturing technique.  Specimen y is large, asymmetrical, and made from white quartz.
 

    All soil was screened through a quarter-inch mesh.  Nevertheless, two items cannot be recorded as to provenience.  One is a piece of micaceous hematite (bb in Figure 5), perhaps a "paint-stone," showing very fine abrasion marks, a smoothed notch at the left end, and several deep parallel grooves, V-shaped in cross-section, at the right end.  The other is a small triangular scraper of white quartz.  No organic remains were recovered from Levels2 and 3.

DISCUSSION

     So much, then, for the recorded material from the Mianus Gorge Rock Shelter.  The question naturally raised is, "What conclusions can legitimately be inferred from such a small collection?"

     A brief review of the work of others in this area clarifies the archeological problems involved and places the Mianus Gorge Rock Shelter in perspective.  Fifty-odd years ago, Harrington excavated rock shelters six to seven miles west of here, as noted (Harrington, 1909).  At least one of these shelters suggested occupancy by different groups, separated by an unknown interval of time.

     At the Finch's Rock House Site, for instance, two layers, subsequently designated Finch I and Finch II by Smith (Smith, 1950;  p. 163) in his synthesis of the archeological manifestations of coastal New York, were found separated by a layer of sterile sand.  The upper layer, Finch II, produced grit-tempered pottery which Smith attributed to the Clasons Point Focus of the East River Aspect.  Contact material  (sheet metal arrowheads) also came from Finch II.  This layer was separated from the lower layer, Finch 1, by an intermediate layer which Smith termed "sterile sand" and is indeed one basis upon which rests the postulation that Finch's Rock House represents a multi-component site.  However, I note in Harrington's original publication that this layer was recorded in one instance as a "narrow band of yellow sand and ashes." The presence of ashes here wants further explanation, and may throw some doubt on the sterility of this intermediate layer.  Further doubt arises from Harrington's own disclosure on pages 128-129 (op. cit.):


 "When the exploration described above had been completed it was thought that everything of value had been found and removed from the cave; but on further deliberation, taking into consideration the darkness of the cave and the blackness and stickiness of the cave dirt, it was thought best to sift the entire contents.  The results were surprising.  The earth had all been carefully trowelled over, then thrown with a shovel so that it could be watched - but a great number of things had been overlooked, as the subsequent sifting showed. Of course all data as to depth and position have been lost, yet the specimens are valuable as having come from the cave…"  (Italics mine.)


     Among other items thus recovered were "95 perfect or nearly perfect arrow points." I suggest that some of this material may have derived from the so-called sterile layer.  This would correspond much more closely to conditions observed at the Mianus Gorge Rock Shelter, where material was separable only on the basis of artificial stratigraphy bolstered by typological analysis.  This does not impugn Finch's Rock House's claim as a multi-component site, but does question the clear-cut separation said to have been evident there.  This condition was not noted at the Mianus Gorge Rock Shelter.  Be that as it may, Finch I contained no pottery, but did have an inventory of rude, chipped, quartz specimens.

     Harrington also investigated several other rock shelters in the region, none as extensive as Finch's Rock House.  Most of the deposits in these shelters were quite shallow; several contained contact material and miscellaneous pot-sherds.5

     Who were the first occupants of Mianus Gorge Rock Shelter?  The presence of Archaic populations in Connecticut is suggested by the work of Glynn (1953).  Archaic occupancy of a Connecticut rock shelter can be inferred from an early report by MacCurdy (1914).  In the region in question, the lower level of Finch's Rock House has been designated an Archaic manifestation (Smith, 1950; p. 163).  At least one researcher believes evidence sufficient to establish a Finch Focus (MacNeish, 1952; p. 54).

     The evidence from the Mianus Gorge Rock Shelter strengthens the hypothesis for presence of pre-ceramic or Archaic peoples in this area.  Whether this was early or late Archaic is hard to say.  The lower level at the E. D. Prey Site, 90-odd miles northeast of here near East Killingly, Conn., has been characterized by Byers (1959a; p. 240) as of an "…extremely Early Archaic character, perhaps bordering on Unspecialized Lithic." I see faint similarities in his descriptions of that site and conditions at the Mianus Gorge Rock Shelter.  For instance, his obvious reluctance at times to classify many of the "things" he found as even artifacts is reminiscent of an impression I have of some material from the lower parts of the Mianus Gorge Rock Shelter.  In fact, Byers' (1958; p. 1) statement that "Seeing artifacts in some of these pieces of quartz takes almost as much imagination as seeing people and animals in clouds!" describes very well the reaction when examining a mass of chips and forms - many of which might have been used for almost anything by early experimenters in an undefined lithic technology.  Perhaps this same thing plagues others working in the Coastal Northeast, who feel that established types in archeology are sometimes greatly taxed by certain material... (?).

     Thus, a presumed Archaic occupancy at Mianus Gorge Rock Shelter might have been very early.  It may just as well have been rather late.

     Lacking altogether are any polished stone artifacts - usually identified with the Eastern Archaic, though Smith (op. cit.; p. 143) may suggest otherwise.  Interestingly, this same lack of polished stone tools was also the case at the Late Archaic Huntington  Site near Perryville, R. I., mentioned by Byers (1958; p. 6) and used by him to fix upper E. D. Prey in time.  He stated that the Huntington Site was known to collectors many years before he visited it, but there was no record of polished stone tools ever having been recovered there, nor did he find any in his work.  He offers no explanation for this, but concludes that it is unlikely that all polished stone tools could have been removed from the site.

     If it is hard to say when in the Archaic this shelter was utilized, it nonetheless seems logical to reason that it was.  I base this reasoning mainly on (1) lack of pottery in Levels 2 and 3 (which is one reason for their combination in this report), and (2) the projectile point types and associated scrapers and flakes.

(Continued)