level of the present surface of the plain, and the lowest portion of this bed was a couple of feet still lower. In the case of Casale Zaffanella there was no mound visible, but on examination the remains of the settlement were found to be disposed precisely similar to those of the terremare above ground, only the mound was completely submerged in a sea of hardened mud. The explanation of this will be readily perceived when we remember that the amount of submergence respectively shown in these instances is in the inverse ratio to their distance from the lower parts of the plain and its great water-channels. The yearly inundations of the Po and its tributaries extend far and wide, each adding a film of mud to previous deposits. Thus, in the course of ages, the surface of the plain has become considerably elevated. The increase of silt since the terramara settlement of Casale Zaffanella was founded, amounts to 12^ feet a depth sufficient to cover the highest part of the mound. It is difficult to say how much this levelling-up process is accountable for the scarcity of these stations in the lower parts of the Po Valley. That they existed, however, in close proximity to the river is amply proved by the stations at Viadana, as well as one or two others ; for example, that at Brescello in the Parmensian district.
The discovery of the terramara at Taranto has opened up questions of far-reaching significance by bringing into focus the products of foreign civilisations, and thus supplying data which may help to bring these contemporary civilisations within the pale of absolute chronology.
Extent.— As to the actual dimensions of the terramara settlements it is difficult to procure accurate measurements, for several reasons. In many instances they are either built over by modern buildings, or there is nothing to distinguish their débris from the surrounding soil without making extensive excavations. Even when the site is a clearly defined mound, as at Montale, one estimate may differ from another, according as the area of the surrounding dyke is or is not included in the measurements. Generally speaking they are quadrangular in form, and, according to Chierici (B. 105, p. 105), their average superficial area is tre ettare (about seven acres). But their respective areas vary very much, as will be seen from the
