The burrow system, or den, in which spectabilis stores its caches of food materials, has its nest, and remains throughout the hours of daylight is a complicated labyrinth of tunnels. Ejection of refuse and soil from this retreat builds up the mound frequently referred to. These mounds are, as Bailey says, characteristic of the species, and are as unmistakable as muskrat houses or beaver dams, and as carefully planned and built for as definite a purpose—home and shelter. They are, furthermore, the most notable of all kangaroo rat dwelling places (Nelson, 1918, 400). They range in height from 6 inches to approximately 4 feet and from 5 to 15 feet in diameter.

The mound is built up not only through the cleaning out of chaff and other food refuse, but through extension and modification of the tunnels; old tunnels, entrances, and caches of musty food material are from time to time closed up and others excavated, repair and rebuilding being especially necessary after the collapse of portions of the den as a result of heavy rains or trampling by cattle. Ejected material is most commonly simply thrown out fan-wise from the openings, without much apparent effort to add to the height of the mound.

There are usually from 6 to 12 entrance holes in each mound opening into the subterranean burrow system, each hole from 4 to 5-1/2 inches in diameter. These holes are nearly all situated a little above the surface of the surrounding soil, and as Price has suggested (in Allen, 1895, 213), this is doubtless a wise provision against flooding, as torrential rains sometimes occur in the kangaroo rat country.

Both Bailey and Nelson state that as a rule several of the holes are closed with sand or miscellaneous earth and old storage material during the daytime, but our observations on the Range Reserve are that such closing is only occasional. Many occupied dens have not a single opening closed. Further, night observations disclose that the inhabitant of the mound will appear from some one of the two or three most-used openings when night falls, and not necessarily from one which has been closed by day. Recently an opening closed one day was observed in use during the night, but was left open all the next day.

In attempting to determine whether there exist similarities of plan or system in the dens, it was considered advisable to map them with some degree of accuracy. This we were enabled to do by laying off a square about a given mound, 2-1/2 or 3 meters each way, and subdividing it into a series of small squares of half a meter on each side by drawing cross-lines on the surface of the ground over the top of the mound. One person then did the digging and exploring of the tunnels, as to direction and depth, while the other noted the results on coordinate paper (Figs. 2 and 3); the proper excavation and mapping of one of these workings occupied from four to eight hours for the two.

While there is greater complexity in the larger, and probably older, mounds than in the smaller, all are extremely complicated and can only be described as labyrinthine in character. The tunnels wind about through the mound, rising and falling in vertical depth, intercommunicating frequently, but with occasional cul-de-sacs, and in places expanding into chambers, of which there may be three or four large ones. The stored materials are found in some, but not necessarily all, of these chambers, and may also occupy considerable lengths of ordinary tunnel, especially when the quantity present is large. Small evaginations of the tunnels frequently contain lesser caches, and it is in such pockets that bits of fresh material are placed during a growing season, or that grain supplied the previous night for bait is usually found.

The main masses of storage are most often found centrally located at depths of from 15 to 57 centimeters, although at times one may find a cache near the periphery of the system and as near the surface as 2 or 3 centimeters. In the latter case the materials are subject to wetting from rains, and consequent spoilage.

The major portion of the whole tunnel system is within about 50 centimeters of the surface of the mound, but usually some one branch tunnel goes to somewhat greater depth, and this is likely to be the one containing the nest; this is also likely to extend toward or beyond the periphery of the main system, and always ends blindly. Such a one, from which two young were taken on January 31, 1920, was at a depth of about 65 centimeters, and about 1-1/2 meters beyond the periphery of the mound itself.

The individual tunnels average about 8 centimeters in height, and about 11 centimeters in width, though the variation, especially in width, is considerable. The expansions mentioned as being the chief places of storage are from 15 to 25 centimeters in diameter, and may or may not involve a considerable increase in height. They are frequently located at junction points of two or more branches of the tunnel system.

The nest cavity is a chamber of approximately spherical shape and from 17 to 23 centimeters in diameter. Chambers of this character were observed and noted as "old storage" in a number of cases. They were sometimes cut off from the rest of the habitation, and at first were supposed to contain abandoned musty storage. As experience in excavating and interpreting results has been gained we have concluded that these chambers in fact represent abandoned nests.

Bailey gives the dimensions of nest chambers observed in New Mexico as about 6 by 8 inches to 8 by 10 inches. The nest is composed of finer, softer, and more chaffy material than the regular storage. The chaff refuse from the food probably contributes largely to it, though some leaves of grasses not stored for food may also be found, and a nest, especially the one in use, may be distinguished, if excavating is carefully done, by the distinct cavity about the size of a fist in its interior (Pl. IX, Fig. 1). One may sometimes find this cavity distinctly warm from the recent presence of the inhabitant.

The walls or partitions between the chambers and tunnels are in places surprisingly thin, and it is no wonder that one is almost certain to break through in stepping on a mound, since the whole is a honeycomblike structure of from two to four stories in vertical plan, as shown by the transect of a mound in Plate VII, Figure 1. As Bailey writes, these partition walls are a mixture of earth and old food and nest material discarded years ago, resembling the adobe walls of the Mexican houses built of chopped earth and straw. This is the result of the continual ejection of refuse and earth as before mentioned, combined with the caving action of rains and disturbances from larger animals.

Apparently there are no special pockets for deposit of feces in Dipodomys burrows; such matter may be found throughout the den, and is more or less mixed with the food refuse which carpets practically the entire tunnel system. The nest and food stores are, however, clean and neat, the droppings being dry and, though present on the floor of a storage chamber, not actually mingled with the food. Evidently the animal does not clean up the floor litter before storing food material.

The entire system for any one den seems to consist not only of the burrows within the mound itself, as described, but of those small outlying ones which we have referred to as subsidiary burrows. These are two to four in number, and are connected with the main mound by the runways already mentioned. They often seem to be way stations on the runways connecting main mounds, and there is seldom any mound of earth whatever in connection with them. One entire den system, the home mound and three subsidiaries, was mapped after being excavated (Fig. 3), all having been carefully gassed with carbon bisulphide. The subsidiaries were simple and contained no storage. Two of them were shallow, while in the third a depth of 48 centimeters was reached. They appear to be merely places of refuge, though the well-worn trails connecting them with the main mound indicate regular use. These runways are conspicuous on the Range Reserve, and are apparently characteristic of mounds throughout the range of the animal. Dwellers in different mounds must have rather extensive social contacts, notwithstanding the enmity of individuals toward each other in captivity. The main mound, in this instance very complicated, was in one place three stories high, and we have found as many as four utilized stories; but as a rule there are two or three only.

Since collapses are rather frequent during rainy seasons, aside from the trampling previously referred to, the kangaroo rats, where abundant, as on the Range Reserve, may well be a factor in increasing soil porosity and fertility; for in the course of time they probably have succeeded in plowing and cultivating the whole surface layer of the soil. They may thus be a factor in ecologic succession, tending to improve the character of the soil and adapt it to another stage.

Doubtless their own workings afford the only shelter the animals know. In the course of our digging in one mound, the occupant, an adult male, did not forsake the den until the excavation was three-fourths completed; and even then it did not leave by a burrow leading away from our operations, but came toward us, escaped the active efforts of four individuals bent on its capture, and ran speedily along a used runway toward another burrow several meters distant. A sack had been stuffed in the mouth of this, however, and, baffled, the rat then returned to the original burrow and was captured. Observations on other rats thus driven from the home mound indicate that they are very familiar with the runways of the vicinity of the mound and the various subsidiary burrows, and it is a question whether they need to see clearly to follow these runs. Apparently they never attempt to escape by forsaking their well-traveled runways. Tests of the maze-running ability of these animals by animal-behavior experts would be of extraordinary interest, in view of the character of the homes which they always inhabit and the network of runs on the outside.