In two previous publications (bibl. 1 and 2) I have brought the formation of calderas into relation with the gas phase, observed by Perret during the eruption of Vesuvius in 1906 (bibl. 3). In these papers I arrived at the conclusion that during the gas phase a cylinder is cored out, and that this may be the cause of caldera formation. In the first paper the subject was treated geometrically, while in the second calculations were made of a particular case (the Krakatoa eruption of 1883) to see if they would bear out this theory. This caldera-formation, however, is not a typical case, as there must previously have been an older Krakatoa-caldera, and in Aug. 1883 it was not a large portion of the volcanic cone that disappeared, but only an island which projected little above sealevel; the northern part of the ancient island Rakata, with the volcanoes Perboewatan and Danan. How a caldera might be formed from a cored-out cylinder I have tried to explain in two different ways. In the case of the Tengger-caldera I assumed, in analogy with what happened in Vesuvius after 1906 (bibl. 3 and 4) that the uppermost part of the cylinder was transformed into a funnel-shape by crumbling away of the walls, and that rising lava, as in Vesuvius 1913—1926, formed a flat bottom which continually reached higher levels. This explanation does not apply to the caldera of Krakatoa, as after the great eruption of Aug. 26th to 28th 1883 no further signs of eruption were observed, until in Dec. 1927 a new phase began in this famous volcano. In the case of Krakatoa in 1883, therefore, I thought it justifiable to apply the phenomena, known to occur in coal mining, of recent subsidences which are caused by the working of coal seams lower down.