424 YOUNG, ERICKSON, AND FAIRHALL bottom of the box. The motor in the suction unit operates from a 12- volt automobile battery or, by means of a step-down transformer built into the box, 115-volt alternating current. Typical sampling times to collect sufficient CO, for analysis, about 5 liters, range from 45 to 90 min. The reason for this range in time is that water is preferentially adsorbed by the sieve and displaces CQ . During times of the year when the total water content of the air is high, collection times have to be reduced to avoid displacing CO,. This also reduces the total amount of CO, collected; however, a sufficient volume is obtained for counting. Studies of various sections of the cans reveal that CO, is quantitatively removed from the air, provided the collec- tion time is not too long. This has been checked independently by com- paring the amounts of water and CO, which outgassed from the exposed sieve material. Atmospheric CO, concentrations are known to be about 315 ppm, and the water concentration of the air that is sampled can be calculated from the relative humidity. The calculated ratio of water to CO, in the air that was sampled agreed with the amounts of water and CO, that were recovered to within about 10%, which is the limit imposed by the memoryof the sieve for water. Since CO, is quantitatively removed, it can be used to calculate the quantity of air that was sampled. We have used this technique to calibrate airflow through canisters in connection with our aircraft sampling program of fallout debris in the lower stratosphere. The molecular sieve produces considerable quantities of radon during outgassing of the sample. Our standard procedure is to outgas the sieve, convert the CO, into methane for counting purposes, and then let the radon decay for four to six weeks before counting the methane. The conversion procedure removes any trace gases (e.g., NO, and SO,) that might affect the counting characteristics of the gas. The double-top paint cans are sufficiently sturdy that they can easily be shipped to remote locations without damage or leakage. The only precaution to be taken is that, when shipments are made byair, the shipping container must provide sufficient pressure to the ends of the cans to keep the reduced external pressure from popping the lids. Likewise, when sampling in extremely cold climates, the cans should be warmed to approximately room temperature before they are tightly sealed. Brief exposure to air in the laboratory during normal handling procedures does not significantly contaminate the sample. Routine collection of atmospheric CO, samples with this type of device has been made in the Puget Sound area (48°30’N, 122°W) ona weekly basis since September 1963. Samples have been collected about every two weeks at several sites in the Arctic basin since January 1964. Two of the Arctic sites are Point Barrow, Alaska (71°20’N, 156°39’W) and floating ice island Arlis Il. These samples have been collected through the cooperation of personnel connected with the