Figure 100. Blistering of an aqueous solution of Mw = 3.8 × 106 g cm−3 PEO solution at a concentration of c = 2000 ppm. Reprinted with permission from [515], copyright 2005 by the American Institute of Physics. by [515]. The first image shown in figure 100 is a closeup of a single, highly stretched polymer thread such as the one seen suspended between the two rightmost drops in figure 97. At first, the thread is subject to what appears to be a linear instability with a well-defined wavelength [553]. Interestingly, the process goes through several stages, leading to up to 4 generations of droplets of decreasing size (cf figure 100). The phenomenon appears not to be described by the FENEP model (314)–(316), as shown in [548]. Namely, according to (314)–(316), when polymers become fully stretched, the thread breaks in a localized fashion, once more described by a similarity solution, whose exact form depends on the polymer model used [554, 555]. A similar phenomenon of ‘iterated stretching’ had been observed earlier in [556]. However, the theoretical analysis is based on dilute, infinitely extensible polymers, and occurs only in the corner region, where the thread meets the drop. This is contradicted both by experiment (since the instability grows in the interior of the thread, cf figure 100 [515]), and subsequent slender-body analyses of the Oldroyd-B equations [316,548,557]. Finally, in his classical monograph Boys [558] shows a picture of a thread taken from a spider web, showing droplets of various sizes on the thread, whose appearance is strikingly similar to figure 100. The ‘sticky’ droplets have an important biological function in capturing insects [559]. It remains to be seen whether they are produced by the same blistering instability seen in figure 100, or whether they come from a coat of sticky liquid applied to an existing thread, as is usually supposed in the literature [559]. 6.2. Shear-thinning fluids If the polymer molecules in solution become stiffer, ideally like rods, memory effects diminish. A particularly simple case is the ‘generalized Newtonian liquid’ [529], for which the stress tensor is proportional to the deformation rate tensor D (cf (39)), but with a viscosity that depends on the invariants of D. Very often it is observed that the viscosity diminishes like a power law as a function of the second invariant ˙γ = DijDij , if ˙γ is greater than a critical value ˙γ0 (Carreau fluid): η/η0 = (1 − β)[1 + (γ /˙ γ˙0) 2 ] (n−1)/2 + β. (320) For low shear, the viscosity tends towards a constant value. If n < 1, as is observed for xanthan gum solutions [560], the behaviour is shear thinning. Note that (320) implies the same shear-thinning behaviour under extension; namely, in a thin filament ˙γ = |∂vz/∂z| = |v | to leading order. This determines the slender-jet equations (30), with the viscosity replaced by (320). The problem is that there is not much evidence so far that (320) is a faithful description for extensional flows, as most rheological measurements are under shear. In fact, there is evidence that xanthan gum solutions (like suspensions of rigid rods), thicken under extension [561–563]. By contrast, there is some evidence that concentrated suspensions of spherical particles show not only shear thinning [564] but also extensional thinning [565]. The latter has also been found in polymer melts [566]. It is simplest to analyse pinching in the limit ˙γ γ˙0, for which there is pure power-law behaviour, and a similarity solution of the form (199) can be found [567–569]. Namely, from (29) one estimates that v ∝ t−1, and a capillary-viscous balance gives 1/h ∝ ηv ∝ vn. From this we deduce immediately that α1 = n: not surprisingly, pinching is speeded up, because the viscosity decreases as pinching progresses. As in the previous section, pinching experiments have the potential of conveying useful information on the extensional rheology of the liquid, which is difficult to obtain by other means. This statement is not tied to power-law behaviour, although this of course simplifies the analysis. The axial extension of the solution is determined from including inertia in the balance [567, 568], leading to α1 = n, α2 = −n/2, β = 1 − n/2. (321) Thus as n decreases, the solution becomes more localized, and for n = 2/3 axial and radial length scales become the same, so that a long-wavelength description is no longer justified. This makes intuitive sense, since the dynamics accelerates rapidly as one gets close to the pinch point. In addition, as solutions of the similarity equations demonstrate [568], profiles become increasingly symmetrical. The net result is that even if the base viscosity is high, thread-like features close to breakup disappear, as seen in the one- and two-dimensional simulations shown in [522]. Instead, breakup is localized and less likely to produce satellite drops. From an engineering point of view, this is a desired effect for many foods, for example yoghurt, for which a ‘stringy’ appearance is not considered attractive [570]. In fact, preliminary experiments show that some brands of yoghurt are indeed shear thinning. Breakup experiments then demonstrate that the pinching exponent α1 is indeed smaller than 1, showing that the thinning effect extends to extensional flow [570]. If the base viscosity is high, one expects inertia to be irrelevant initially; of course, as viscosity decreases, a crossover to an inertial–viscous balance will occur even sooner [567,568]. For n = 1 new branches ofsymmetric solutions are 70