I42 REPORTS ON INVESTIGATIONS AND PROJECTS. 



off the planulae, which are very small, a bag of fine-mesh bolting-cloth must 

 be affixed to any tube used in withdrawing the stale water. One method was 

 to siphon off the stale water with a rubber tube, the end of the tube inserted 

 into the culture-jar having been drawn over one end of a glass tube, the other 

 end of which is enveloped in a bolting-cloth bag. The table on which the 

 culture-jars stand is provided with a gutter into which the water drawn off 

 is discharged, ultimately flowing outside the building through a pipe through 

 the floor. After a jar has been emptied to within an inch of the tile, it is 

 refilled with fresh sea-water. This method causes a change in the level of 

 the water, and by the pouring stirs up the unattached planulae. A second 

 method was to withdraw the old water by a glass siphon resting on the upper 

 edge of the jar, the siphon having been rendered non-emptying by having its 

 outer end bent upward. Fresh sea-water is added by a siphon extending to 

 the bottom of the culture- jar from a supply-jar placed at a higher level. By 

 this method a constant level is maintained in the culture-jars; the old water 

 is drawn off from the top while the new water is added at the bottom. A 

 third method was to have inside the culture-jar a tantalus siphon emptying 

 through the side of the jar near its bottom. Fresh water is siphoned into the 

 culture- jar from supply- jars placed at a higher level. When the water in a 

 culture-jar has reached the level of the upper curvature of the siphon, it 

 begins to run out and continues to flow until the level of the open end of the 

 siphon in the jar is reached. The jar is then refilled by the afferent siphon 

 until the level of the upper curvature of the tantalus siphon is again reached, 

 when the water again begins to flow out. This method causes a rise and fall 

 in the level of the water. A fourth method was to cut the bottom out of a 

 culture-jar and to place the glass collar thus produced over a tile in a jar of 

 larger diameter, the bottom of which had previously been covered with sand 

 to a depth of an inch or slightly more. The tile and its surrounding collar 

 are sunk into the sand until the upper surface of the tile and the upper sur- 

 face of the sand are level with each other, while the level of the upper edge 

 of the collar remains slightly higher than that of the inclosing jar. Water 

 is siphoned into the collar from supply- jars, and filters through the sand 

 filling the space between the collar and the side of the inclosing jar. When 

 the level of the upper edge of the jar is reached, the water overflows. This 

 method maintains a constant level of water, draws off old water at the bot- 

 tom, and adds new water at the top. 



As all four of these methods were tried, and as all were successful, the 

 preference between them was not determined. Fresh water is necessary, and 

 occasional stirring of unattached planulse may be beneficial. 



The two previous attempts to carry attached larvse over from one year to 

 the next failed. As it was discovered in 1908 that attached larvse thrived in 

 a floating live-car, it was decided to try to build a live-car strong enough to 

 weather a hurricane, to load it with tiles bearing young polyps, and to anchor 

 it securely. Two views of this live-car are here presented. One, plate 1, 

 fig. C, shows its structure and the method of fastening the tiles ; the other, 

 plate 1, fig. D, shows the car after being placed in the water. Buoyancy was 

 obtained by a closed copper tank in each end. The car is made fast to a sea- 

 anchor by an anchor chain.* 



*This live-car was sunk just before the hurricane of October 17-19, 1910, but was re- 

 floated, and, according to a report received from John Mills, the head mechanic of the 

 laboratory, 6 tiles bore a total of 32 living polyps, or one-fifth of the polyps had sur- 

 vived the sinking of the car and the hurricane. 



