Two model systems were used to study the (2,5,9). This principle has been applied to studying the freezing ofwater in soils and building materials. Both the freezing temperature and the spread of ice were influenced by the size and distribution of small diameter pores (2,(5)(6)(7). Calorimetry has been used to determine the melting point of water in porous materials so that the porosity and pore size distribution can be estimated (5-7). Possible effects of small diameter pores on the freezing of water in plant tissues have received little attention. Mazur (9) discussed this principle and related it to the prevention of ice propagation through cell membrane pores and into the protoplasm. George (4) recently observed that melting began near -30C in water saturated sycamore heartwood. He speculated that this fraction of water may be located in cell wall microcapillaries and discussed the possible role of the microcapillaries in limiting the spread of ice in these tissues. In reduced to 16 mm to facilitate temperature equilibration between the chamber and cooling bath. The chamber was fitted with 13-mm diameter polycarbonate filters (Nuclepore Corporation, Pleasanton, CA 94566) or aluminum discs. The chamber was cooled in a 30% ethanol bath held at -12C. Chamber temperatures were monitored using copper-constantan thermocouples and a recording potentiometer. The experimental apparatus is illustrated in Figure 1. Chamber 1 was designated as the ice donor chamber and was filled with a suspension of an ice nucleating strain of Erwinia herbicola (Lohnis) Dye, isolate 26 (106 cells/ml). Nuclepore filters or aluminum discs separated the donor from the second chamber. Thermocouples were inserted into both sides of the modified Boyden chamber which was then lowered into the -12C bath. Freezing was initiated in the donor chamber between 0 and -1C. The temperature at which freezing was initiated in the second chamber was recorded. Individual experiments required 15 min and were repeated three times. Values reported are the mean plus or minus the standard deviation.Experiments with Controlled Pore Glass. Controlled pore glass particles (Sigma Chemical Co.) were obtained with a range of nanometer pore diameters (7.5 ± 6%, 16 ± 12%, 33 ± 8%, 73 ± 9%, 149 ± 6%, and 300 ± 9%). Glass particles were uniformly sized (80-120 mesh) by the manufacturer. Both the particles and the pores were irregularly shaped as observed using scanning electron microscopy (data not presented). Dry glass particles were placed into modified Beem capsules (Ladd Research Industries, Inc., Burlington, VT 05402) (100 psl) and saturated with either deionized water or dilute aqueous solutions.The freezing of water within saturated controlled pore glass particles was characterized using DTA.2 The technique was a modification of that described by Quamme et al. (11). The junction of a 40-gauge copper-constantan thermocouple was submerged into the slurry of glass particles. Excess solution was removed using a micropipette. Dry controlled pore glass was ...