Snomax® is often used as a surrogate for biological ice nucleating particles (INPs) and has recently been proposed as an INP standard for evaluating ice nucleation methods. We have found the immersion freezing properties of Snomax particles to be substantially unstable, observing a loss of ice nucleation ability over months of repeated droplet freezing measurements of the same batch of Snomax stored as dry pellets in a freezer. This reflects the fragility of the most ice active large protein aggregates and presents issues for the use of Snomax as an INP standard. The ice nucleation properties we determined using a fresh Snomax batch agreed well with the recent method intercomparison from the Ice Nucleation UnIT (UNIT) project, while an older batch did not. Using an oil immersion droplet freezing technique, repeated freezes of Snomax droplets resulted in a decrease in ice nucleation ability after successive refreezes. We attribute this to the disruption or displacement of the most ice active protein aggregates that are thought to contain the ice nucleants. Partitioning of the protein aggregates from the droplet into the immersion oil that is accelerated by droplet freezing events could explain the observed decrease in freezing ability. Droplets in mineral oil or low viscosity silicone oil experienced a smaller reduction in freezing temperature than when squalene oil was used. The effect of the immersion oil may be specific to proteinaceous biological particles, and we have not observed it in nonproteinaceous materials. Caution is warranted in the use of oil immersion droplet freezing methods to determine immersion freezing properties.