210 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS tests is based on anecdotal evidence and opinion regarding how long a product should take to reduce the numbers of the challenge inocula. The best way of interpreting the data, however, is to compare how the test product performs against how well-preserved and poorly preserved control products perform. Well-preserved products are those that do not become contaminated during consumer use, and poorly preserved products are those that do become contaminated when used by consumers. OTHER PET METHODS D-value methods. Rapid tests are sometimes used for quick impressions of which preser- vatives to use in a product. One such method is the D-value method. Aside from one author, no one else claims D-value methods are valid for final testing of nationally distributed product (114). In fact, D-value methods are inappropriate for at least some consumer products (115). This method is actually an adaptation from food microbiol- ogy's heat or radiation destruction D-values. Heat and radiation kills do indeed follow first-order rate kinetics, and therefore the D-values determined for them are quite valid. However, biocide kills follow second- order rate kinetics (115,116). The only case where a second-order reaction can approach pseudo-first-order rate kinetics is when the second reactant (biocide) is present in such large excess that it is virtually in constant concentration. Preserved products do not have an excess of preservative such that the biocide remains in constant concentration when contamination occurs (117). A biocide-organism reaction is stoichiometric the biocide does not act like an enzyme that catalyzes a reaction where live organism goes to dead organism, but the biocide is not spent. Therefore, since the biocide-organism reaction is second order, with the biocide serving as the limiting reactant, D-value tests based on first-order rate kinetics are invalid (115,117). The second criticism of the D-value technique is that it extrapolates beyond the mea- sured data by falsely assuming a linear relationship between biocide exposure time and the number of surviving microorganisms. In defense of rapid D-value methods, how- ever, one may find they allow a preliminary screening of preservatives. This approach assumes that appropriate reproducible controls are in place such that one will be able to rank the various D-values for a wide variety of products and be able to correlate that data to full-scale PET results on the same products. Capacity tests. A capacity test determines how many bacterial challenges are needed before the product begins growing microorganisms (118). After each challenge, the products are sampled and challenged again until the product either receives 15 chal- lenges without showing growth (a well-preserved product) or until three consecutive positive results occur (a lesser-preserved product). The goal is for the product to reduce the number of viable organisms by 3 logs (99.9%) in 48 hours. With each subsequent challenge, this ability diminishes as a result of dilution, neutralization, and reaction with the inoculum. The claim, by some studies, that multiple challenges provide no more information than single challenges (67) may actually be more pragmatically based than scientifically based. The reason why multiple challenges with low levels of organ- isms are not the same as one challenge with a high level is similar to the concept of the Danysz phenomenon in immunology (119), where when a high level of inoculum is used, the biocide combines with an equivalent amount of microbes, allowing the challenge to be killed, but when challenging multiple times, each challenge combines

COSMETIC PRESERVATION 211 with more than its equivalent amount of biocide, leaving insufficient biocide to react with subsequently added microbes in the challenge (120). The value in such a test may be for multi-use products. However, for a more predictive test of consumer contami- nation potential, one should lower the challenge inocula to levels likely to be encoun- tered during use. Once done, the capacity test may be a quantitative and valid way to understand a product's ability to handle contamination from use. PRESERVATIVES AVAILABLE FOR USE MODE OF ACTION OF PRESERVATIVES The mode of action of antibiotics is known at the molecular level since they act via specific biochemical reactions. In contrast, the modes of action of preservatives and biocides are far more generalized,with numerous points of attack. Nearly all biocides work by denaturing cellular proteins or by affecting membrane permeability so that either transport or energy generation is blocked. For example, chlorine oxidizes reduced sites of organic compounds, including proteins, throughout the bacterial cell. Protein denaturants also include formaldehyde, formaldehyde releasers, isothiazolinones, and bromine compounds. The parabens and weak acids (e.g., sorbic, benzoic, and dehydroacetic acids) disrupt control of membrane electrical potential to block energy generation and nutrient trans- port (121). Thus the parabens apparently inhibit nutrient uptake by shutting down permeases, disrupting porin channels, or by disrupting the membrane pH gradient or electrical charge potential across the membrane to prevent substrate transport and ATP generation. This inhibition is apparently reversible and is consistent with other obser- vations that the mode of action of parabens is by disruption of the membrane electrical potential (122). Organic acids probably work in the same fashion (123) however, they may even be enzyme inhibitors as well (124,125). Typically, they are only biocidal at pH values below their pK•. In this protonated form, they pass through the membrane, and the hydrogen ion dissociates from the weak acid to decrease the cytoplasmic pH. As a result, both substrate transport and oxidative phosphorylation are uncoupled from the electron transport system. This effectively starves the cell of needed substrate and energy derived from ATP synthetase driven by hydrogen ions. Phenolics disrupt the proton motive force of the cell membrane (126,127). They also have the ability to non-specifically denature cytoplasm, cell walls, and cell membranes (128). The more lipophilic phenolics have the greater antibacterial capacity perhaps because of a greater ability to partition out of the water phase and into the lipid membrane (129,130). Alcohols likewise disrupt the membrane, causing permeability loss (131), and they also appear to inhibit enzymes (132). Perhaps some of the most widely used of the newer preservatives are the isothiazolinones. These are usually compounded into a single product composed of chloromethyl- isothiazolinone and methyl-isothiazolinone, but they can also include benzyl-type com- pounds (133,134). The isothiazolinones inhibit glucose oxidation and active transport without significantly affecting membrane integrity (135). In fact, these compounds denature enzymes and other proteins containing thiol groups (e.g., ATPase, glyceral-