In the last few years, adsorbed water films on soil particles have been the focus of sustained attention in the scientific literature. A number of articles have attempted to improve our understanding and predictions of the transport of water in unsaturated geological media by taking into account the flow that may be occurring in these films [e.g., Tokunaga, 2009; Lebeau and Konrad, 2010]. In a different arena, the significance of straining by adsorbed water films to the retardation of (bio)colloids in soils has been an object of debate among researchers [Crist et al., 2004; Bradford and Torkzaban, 2008]. In particular, it is still uncertain at this point to what extent water films in unsaturated soils or sediments are thick enough to strain and immobilize engineered nanoparticles, given the latter's tendency under field conditions to rapidly form aggregates with diameters in the hundreds of nanometers or even microns [e.g., Baveye and Laba, 2008; Chen et al., 2008; French et al., 2009].
 In this context, the recent article by Tokunaga  presents undeniable interest in that it proposes a detailed description of a theoretical method to evaluate the thickness of adsorbed water films. If its predictions were somehow confirmed experimentally, Tokunaga's  model could be very useful to deal with situations where water films might have a significant influence on water and chemicals movement. However, at this juncture, it is largely unclear whether the model proposed by Tokunaga  provides a reasonable depiction of reality.
 Indeed, one of the key limitations of Tokunaga's  modeling effort is that there are extremely few data that can be used to assess whether or not the calculations that are presented are in the right ballpark for common geological media. Many articles present data related to systems that are not directly relevant to unsaturated soils, such as wetting films containing surfactants [Hänni-Ciunel et al., 2009], adsorbed to polyelectrolyte-coated surfaces [Ciunel et al., 2005], or on highly hydroxylated silicon [Anderson and Ashurst, 2009]. The experimental data that are relevant to soils, on the other hand, are few and far between, and they are not convergent. The 43-year-old article ofRead and Kitchener , which Tokunaga  cites, describes measurements of the thickness of “equilibrium wetting films” of dilute solutions of KCl, LiCl, BaCl2, and LaCl2 on a polished vitreous silica surface. These films are not the result of a traditional adsorption process. Rather, they are produced ingeniously by trapping some solution between a gas bubble and the silica surface. For an internal pressure in the bubbles set by interpolation at 145 Pa, these authors found wetting film thicknesses ranging from 30 to >120 nm. These numbers are large compared to those found in a number of more recent articles, which Tokunaga  does not cite, even though they seem very relevant. In the first of these articles by Beaglehole et al. , adsorption of water on molecularly smooth mica was monitored by using an angle-averaging, refractive-index-matching ellipsometric technique. Near saturation in the vapor phase, the water film on the mica tended to a finite thickness of ∼2 nm, equivalent to roughly seven water layers. In the article byAsay and Kim , attenuated total reflection (ATR)-infrared spectroscopy was used to determine the molecular configuration of water adsorbed on a hydrophilic silicon oxide surface at room temperature, as a function of the amount of water vapor in the air. At a relative humidity as close to 100% as these authors could get, the adsorbed water film had a thickness near 2.8 nm, equivalent to about 10 water layers. In a more recent article using synchrotron X-ray reflectivity (SXR),Bohr et al.  measured the thickness of the water film that adsorbs on a cleavage surface of calcite (CaCO3) in a sample chamber where relative humidity could be controlled within the range from <4% to 90%. The water films that developed in this case were remarkably constant in thickness at 1.55 nm (±0.1 nm) and independent of humidity. The fact that techniques used by Beaglehole et al. , Asay and Kim , and Bohr et al.  are much more sophisticated than those implemented by Read and Kitchener , but especially the fact that more recent measurements were made under conditions similar to what one would expect to find in unsaturated subsurface materials, would appear to lend strong credence to the notion that adsorbed water films on subsurface materials are likely to be very thin. Circumstantial evidence pointing in the same direction is also provided in a very interesting article by Mokady and Bresler . These authors pointed out that to explain the apparent sharp decrease in cation exchange capacity, observed as a result of a decrease in moisture content in a sandy soil, adsorbed water films would have to be very thin, to the point that even molecular diffusion of cations would be “subdued” in these films.
 This relative paucity of experimental data suggests clearly that further detailed measurements on a variety of mineral surfaces, and if possible using a variety of experimental techniques, are needed to determine the typical thickness of adsorbed water films in unsaturated soils and sediments. Yet, at this juncture, what we know of the effect of solid surfaces on the structure of vicinal water raises serious doubt about the large film thicknesses observed by Read and Kitchener  and implied by Tokunaga's  model calculations. Physically, one expects that under equilibrium conditions, the thickness of an adsorbed water film would not be drastically different from the distance at which the underlying solid surface ceases to exert any physical influence on water molecules, either directly or indirectly (through electrostatic interaction with exchangeable ions). There was a time, especially in the prime of the polywater episode [Franks, 1981], when a number of researchers thought that charged surfaces, such as those of clay minerals, could affect the geometrical structure of water molecules over distances equivalent to the thickness of many tens of water layers. However, that perspective did not survive experimental scrutiny for very long. Evidence from many different sources (including Asay and Kim  and Bohr et al. ) have shown that when spectroscopic measurements are interpreted correctly, i.e., differentially and not as integral values [e.g., Sposito and Prost, 1982], solid-water interfaces are observed to influence water properties significantly only over the first 2–3 water layers away from the interface, i.e., up to a distance of at most ∼0.9 nm. A wide range of experimental techniques have been adopted in the research leading to that conclusion. For example,McBride and Baveye reached it unequivocally by monitoring the molecular behavior of a spin probe via electron paramagnetic resonance in saturated pastes of the clay mineral hectorite, characterized by a high-surface charge density. Therefore, on the basis of this evidence, and as long as particle surfaces are not too convoluted and heterogeneous, one would anticipate finding adsorbed films with thicknesses on the order of 1–2 nm on charged surfaces in subsurface environments. One expects thinner adsorbed water films to be prevalent for sediment particles large enough to be covered by adsorbed water films under unsaturated conditions, and which typically have very low-surface charge densities.
 Incidentally, a parallel situation is observed in the case of air-water interfaces, once also believed to exert a long-range patterning effect on water molecules inside the liquid phase [Jungwirth, 2011]. In spite of experimental evidence contradicting this perspective, it has never entirely been put to rest. However, this may finally be changing, as suggested by Jungwirth . A recent, very elegant experimental study by Stiopkin et al.  demonstrates conclusively, as Jungwirth  puts it, “that only the surface layer is distinctly different in structure from the rest of the liquid, and that the water surface has the thickness of just one layer of water molecules, which is about 0.3 nanometers. The presence of electrolytes may alter that picture somewhat, but intuitively one would not expect the changes to be dramatic.
 If one adopts the perspective that adsorbed water film thicknesses of several tens of nanometers are not realistic on typical geological materials, the fact that film thicknesses predicted by Tokunaga's  modeling exercise are of that magnitude raises a number of questions about the approach used. It seems possible that Tokunaga's predictions are due to the values assumed for some of the parameters in the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory. In particular, application of the DLVO theory to adsorbed water films (assuming such application is indeed sound) requires assumptions to be made about the value of the Hamaker constantA132, which is extremely difficult to measure directly. In very specific cases, the microscopic approach of London and Hamaker, or the more rigorous, macroscopic Dzyaloshinskii-Lifshitz-Pitaevskii (DLP) method can be used to calculate Hamaker constants from first principles [see, e.g.,Israelachvili, 1991; Hiemenz and Rajagopalan, 1997; Saramago, 2010]. Unfortunately, computations are virtually intractable in most systems of practical interest; and in systems as simple as molecularly smooth mica, evidence suggests that the DLP method is not applicable [Beaglehole et al., 1991]. Under these conditions, the Hamaker constant is therefore generally taken as an adjustable (or “fudge”) factor one manipulates to make the van der Waals interactions come out approximately right to fit experimental data. Alternatively, one can estimate it using an equation for the film thickness on planar surfaces due to the findings of Iwamatsu and Horii , by making the (largely unwarranted) assumption that under very dry conditions, the soil volumetric water content divided by the measured specific surface area provides a reasonable estimate of adsorbed film thickness [e.g., Or and Tuller, 1999; Tuller and Or, 2005]. Regardless of the approach adopted, one cannot assume a priori that an approximation relating several independent Hamaker constants, such as the one Tokunaga [2011, equation (2a)] adopts from Israelachvili  to compute A132, would work in one system if it has been shown to work more or less in another. Given all this, at the very least it would be worthwhile to determine how the values assumed for the Hamaker constant A132 influence the final outcome of a modeling exercise such as that proposed by Tokunaga .
 Aside from these uncertainties associated with parameters of the classical DLVO theory, a broader question concerns the extent to which this theory is sound. Even in the situations for which the DLVO theory was originally devised, i.e., the interaction of paired, identical, planar-charged surfaces, fundamental criticisms have been frequently levied against its theoretical underpinnings [e.g.,Langmuir, 1938; McBride and Baveye, 2002, 2003; Ise and Sogami, 2005; Smalley, 2006]. Many examples exist of situations where the DLVO theory fails to account even qualitatively for experimental observations [see, e.g., Chung et al., 2012]. In the past, whenever researchers [e.g., Churaev and Derjaguin, 1985; Christenson, 1988] have reported agreement between DLVO-based predictions and experimental observations, various modifications of the theory, in the form of one or more ad-hoc “non-DLVO” or “extra-DLVO” forces [Christenson, 1988], have most often been required to produce a half-way reasonable fit. In aqueous solutions, these non-DLVO interactions are commonly referred to as “hydration” forces. Unfortunately, as pointed out accurately bySaramago , the exact physical origin of these forces has remained entirely elusive to this day, in spite of protracted efforts to elucidate them. In fact, in many ways, the heuristic tweaking of the DLVO-theory, which non-DLVO forces amount to is reminiscent of the periodic improvements of the Ptolemaic planetary system in another era. In this context, one may wonder whether theories related to the physics of adsorbed wetting films have matured enough yet to serve as a solid foundation for a modeling effort such as that undertaken byTokunaga .
 To summarize this Comment, it appears that Tokunaga  has rendered a very valuable service in pointing out the possible significance of adsorbed water films in unsaturated subsurface environments and suggesting an approach to evaluate their thicknesses. The predictions of his model raise many intriguing questions on what we do and do not know about water films in soils and other geological materials. In particular, if it turns out that adsorbed water films in soils typically are much thinner than currently advocated in a number of models in which relatively thick water films are needed to fit water transport data, there may be something wholly fundamental about the movement and retention of water in unsaturated soils that we are still missing. I hope that this Comment will stimulate interest in direct measurements of water film thicknesses, which appear to be an essential requirement before we proceed further, and in a detailed scrutiny of the various theoretical frameworks available to describe these films. Regardless of the outcome, developments in both of these directions will ultimately result in a better understanding of the behavior of unsaturated soils.