surface was more regular, lacking the jagged cracks and deep fissures of the hilltop outcrop; and it lacked the overhanging projections which provided overhead shelter for the rats along the upper outcrop. More than ninety per cent of the rats that were recorded as associated with the outcrops were at the hilltop stratum.

Second in preference to the hilltop outcrop as a house site was the base of an osage orange tree in thick woods. This tree occurs throughout the woodland of the Reservation, having become established when the leaf canopy was more open, and the whole area was subject to grazing, with less development of the understory vegetation in the woodland. Houses were most often situated in those osage orange trees that had been cut one or more times, and had regenerated with spreading growth form, the multiple branching stems offering substantial support. Occasionally houses were built in crotches from two to six feet above ground.

Blackberry thickets also are favorable locations for houses. These thickets grew up mostly in fenced areas from which livestock were excluded, but where there was not dense shade—hilltop edges and level or gently sloping ground adjacent to creek banks. The houses were usually in densest parts of the thickets where they were almost inaccessible. Mats of dead canes more or less horizontal, with the live canes growing up through them, provided effective overhead protection, while the ground beneath was relatively open. Houses built in the thickets were so well concealed that they were usually not detected until after leaves were shed in autumn. In most cases the blackberry thickets were small and well isolated. Houses of the rats were sometimes unusually near together suggesting that these thickets provided especially favorable habitat conditions.

Hollow trees are often utilized, the accumulation of sticks for the house being largely inside the cavity. To be suitable for a house site, the snag must have an opening near ground level, and another higher on the trunk, providing emergency outlets in two directions. Most of the hollow trees utilized were black oaks (Quercus velutina).

In 1948 there were many houses in cut tops of trees left from small scale lumbering operations a few years earlier. The densely branched tops of elms, oaks and hickories had satisfied the requirement for support of the house and nearby shelter. The houses built in them were in open woodland well separated from otherwise favorable situations. By 1948 the tops were disintegrating and no longer provided effective shelter. The houses built in them were falling into disrepair and were not permanently inhabited but were often used temporarily by wandering individuals.

Along cut banks of gullies where trees were partly undermined by erosion, the exposed, tangled root systems provided sites for occupancy. In these situations the accumulations of sticks were small and lacked the typical domed shape, consisting essentially of a lining to the cavity beneath the roots.

Two small buildings at the Reservation headquarters were accessible to woodrats and were utilized off and on throughout much of the period of this study, despite the fact that most other sites of occupation away from the hilltop outcrops were deserted in the same period. One small building used as a laboratory had an enclosed wooden box five feet square housing an electric water pump. The interior of this box was accessible to the rats from beneath the floor. Litter of sticks and stems and various food materials were carried in by the rats. The nest thus protected and enclosed was not surrounded by the usual accumulation of sticks. An old garage 30 feet from the laboratory building was also occupied, sometimes by a different individual. The nest and food stores were behind boards propped against the wall.

In October, 1948, live-trapping was begun on a heavily wooded slope facing northwest, and a ten-acre area was trapped rather thoroughly in the succeeding weeks. Because few traps were then available, this was the only area that was well sampled in 1948, although diffuse trapping was carried on over some 200 acres. On the ten-acre tract a total of 17 adult and subadult woodrats were caught, four along the hilltop rock outcrop, six along the gully at the bottom of the slope, and seven at intermediate levels on the slope. Judging from the many unoccupied houses, the population on this tract had been much higher before the study was begun. On the basis of this sample it seems that in 1947 a population of several hundred woodrats lived on the wooded parts of the square mile where the Reservation is located.

Reduction of Population

The abrupt reduction in the population of woodrats on the Reservation cannot be explained conclusively with available data. Probably weather played a major part, but other unknown factors must have been important also. It is certain that the population of woodrats was high, if not at an all-time peak, in 1947. In late February, 1948, when one of us (Fitch) first visited the area on a preliminary inspection trip (not concerned primarily with woodrats), houses of these rats were found to be unusually numerous and those seen seemed to be occupied and well repaired. Possibly the population was drastically reduced within the next few weeks, as unseasonably cold and stormy weather occurred in early March. For the first 12 days of March, 1948, temperature averaged 20° below that of average March weather, and even colder than the average for January or February. A reading of -5°F. on March 11 set a new low locally for the month since records were begun in 1869. The record low temperatures were accompanied by 12.8 inches of snow. This spell of unusually severe weather in early March coincided with the period in which first litters of young usually are born, as most females breed in early February and the gestation period is in the neighborhood of five weeks. That most of these first-litter young may have been eliminated by the unfavorable extreme of weather at the most critical stage in the life cycle may be readily imagined although definite proof is lacking. However, the mortality must have extended beyond newborn young. Loss of first litters ordinarily would be compensated for by the end of the season, since a female usually breeds more than once in the course of a season. In any case, by autumn, when the actual field study of woodrats was initiated, many houses were already deserted and in disrepair. Although the rats were still moderately abundant, they were, seemingly, much below the population peak of the preceding year.

Further drastic reduction of adults and subadults took place in the winter of 1948-49. In the course of live-trapping operations from mid-October into early December, 51 individuals were caught and marked. Chiefly because of unfavorable weather conditions, field work was discontinued in mid-December, and live-trapping was not resumed until early March. Subsequently, only 12 of the woodrats previously marked could be recaptured, and the population had become noticeably sparse. Seemingly, more than three-fourths of the population present in late autumn had been eliminated in the interval. In January, weather was exceptionally severe; on the ninth and tenth the worst sleet storm in twelve years occurred. Sleet fell in small granules, while the temperature remained several degrees below freezing. Partial thawing on January 12, 13 and 14 was followed by a steady drizzling rain on the fifteenth. On the following day the temperature dropped to -7°F. Ice still remained from the sleet storm, and the slush again froze. On the night of January 18, there was one of the worst snow storms on record and temperature reached a low of 2°F. Exceptionally low temperatures persisted through January