On Oreology, the fracture and flow of “milk's favorite cookie^®”

Rheology is broadly defined as the study of the flow of materials with complex or non-Newtonian viscosity. Within kitchen-relevant flows, rheology has been used to address diverse challenges in food science¹ in order to rigorously understand and engineer the deformation and flow of these materials at kitchen, laboratory, and industrial scales. For an accessible introduction to practical rheometry, the reader is referred to Ghanbari et al.² Rheology applied to food science has found applications ranging from understanding the structure of cheese using fractional calculus models³ and improving the processing of chocolate to more evenly distribute cocoa butter and thereby enhancing quality⁴ to understanding how flow properties⁵ and microstructure⁶ influence the perceived texture of a wide range of foods. This area of study more recently has expedited the industrial development of new specialty foods, giving a framework to study the composition and flow of gluten-free batter and breads,⁷ modeling the texture of starch thickeners for sauces,⁸ and informing the choice of sweeteners used in low-calorie chocolate to create optimal yield stress and “sensory acceptance” of the sweets,⁹ as the sensory perception of chocolate is heavily influenced by flow and composition.¹⁰ 

For materials with multiple ingredients and time-dependent properties, rheological study may be particularly insightful for understanding and tuning the properties. For example, the flow of Swiss cheese fondue may be reliably tuned by adding starch (corn flour) or ethanol (wine),¹¹ while emulsions like mayonnaise restructure significantly over time after shearing,¹² requiring careful care of their composition and preparation using bulk and interfacial rheology.¹³ 

While rheology cannot address questions like whether beans belong in chili, or whether a spoon or fork is the more appropriate utensil for mac ‘n’ cheese, it can describe why larger grains of cereal or larger nuts rise to the top of mixtures when shaken (by granular convection¹⁴), quantify useful food descriptors like “creamy,” “mushy,” “slippery,” “rubbery,” and “astringent,”^5,15 and pinpoint ingredients and length scales that imbue these textures. Fluid mechanics beyond rheology has also been applied to understand and improve food science,¹ including revealing better methods to dip biscuits into morning tea (horizontally, not vertically) by understanding liquid flow through porous carbohydrate media,¹⁶ showing why microwaving tea is less effective than heating a teapot from below (due to fluid convection),¹⁷ clarifying what causes cold milk droplets to bounce across the surface of hot tea or coffee,¹⁸ and solving how you can carry your nearly-full coffee cup without spilling (i.e., by walking backward).^19,20 At the kitchen sink, interfacial fluid mechanics can describe the changing appearance of bubbles as they drain.²¹ 

Understanding the flow behavior of food ingredients and products has widespread impact on economies of production, public health, and individual diets. Likewise, the experience of consumption (i.e., feel, softness, and texture) of many foods is intrinsic to their enjoyable consumption, with one example being the habit of twisting a sandwich cookie apart before eating. In this study, we seek to address this challenge for our everyday trilayer laminate composite, the Oreo cookie. Although unstudied compared to starches, doughs, and condiments, sandwich cookies present a canonical example of parallel plate rheometry, in which a fluid sample (the “creme”) is initially fixed between two parallel plates (the wafers). When the wafers are counter-rotated, creme torsionally deforms in shear, flows, and ultimately fractures, leading to separation of two cookie parts [Figs. 1(a) and 1(b)]. In a cohesive or bulk failure event, the creme itself will flow until it ruptures and will be distributed between the two cookie wafers. In adhesive or interfacial failure, the creme instead delaminates nearly entirely from the wafer–creme interface at one wafer and remains adhered only to the second wafer. In our tests, adhesive failure is most common when manipulating cookies by hand [Figs. 1(c)–1(e)] or laboratory equipment, with some exceptions when some creme appeared to cling to stochastic asperities on the second wafer.

Importantly, the postmortem creme distribution of sandwich cookies is found to be consistently oriented within a box [Figs. 1(c)–1(e)] for all boxes we tested, as has been suggested earlier.²² During the manufacturing process, Oreos are constructed by dispensing creme onto a first wafer (which we call wafer 1), followed by a second wafer (wafer 2) being placed on top.²³ This difference in initial wafer–creme contact supposedly influences the relative creme–wafer adhesion.²² Due to the high regularity of manufacturing, this results in the first wafer having a slightly higher level of creme adhesion, and this orientation is preserved when Oreos are placed into the final product box. For instance, in a regular-sized box of regular Oreos obtained in Cambridge, MA, wafer 1 predominantly faces toward the left side when the package text is upright, whereas in a “family size” box of regular Oreos, wafer 1 faces to the right side when the package text is upright, both showing 80% of Oreos retaining creme on the wafer facing that preferred side, rather than 50% as would be expected from random chance. We have observed that this in-box preferred orientation varies between flavors and creme “stuf” or creme filling levels, although it is consistent among multiple boxes of a single variety. This challenge prompts the core of our study to understand the influences of cookie breakage kinematics and inherent cookie properties to control failure stress, failure type, and creme distribution and ultimately to robustly identify whether Oreos are—or can be made to be—“fair,” so that they split creme equally between wafers without bias. To what extent is the result predetermined by manufacturing?

While Oreos are not commonly considered to be a fluid, Oreo creme is a member of the class of flowable soft solids known as “yield stress fluids,” which are fluids that act as soft solids when unperturbed and only flow under a sufficiently large applied stress.²⁴ Yield stress fluids include a diverse range of important food materials in addition to Oreo creme such as cookie dough, frosting, ice cream, peanut butter, guacamole, and ketchup as well as foams, cosmetics, biological gels, screen-printing and robocasting inks,²⁵ snow and lava,²⁶ drilling muds (e.g., bentonite dispersions), concrete and mortar. They also include dry but flowing powder ingredients and granular materials,^27,28 from flour and corn starch to collections of sprinkles, marshmallows, fried rice,²⁹ and fire ants.³⁰ (We would be curious as well about how the bite-size, edible protein models developed by Baumer et al.³¹ would roll down an incline.)

Moreover, the measurement of fluid failure behavior into which sandwich cookie separation can be categorized is a broader topic within rheology. In particular, the study of failure of such amorphous (“glassy”) colloidal soft matter and, particularly, stress overshoots and related shear-banding that occur during strain-controlled failure are an active area of experimentation and simulation within soft glassy rheology to understand the underlying physics.^32,33 Further, edge fracture is a persistent challenge in parallel plate rheometry wherein normal stress differences build up by application of torsional shearing strain, causing the fluid meniscus to deform and propagate a radial crack near the outer edge,^34,35 as may happen to initiate the failure of Oreo cookies. While this effect is usually fatal to experimental testing of material properties, it has been productively exploited to rapidly pinch off threads of viscoelastic inks in three-dimensional (3D) printing processes,³⁶ which otherwise leads to stringing defects during printing. In addition, complex failure mechanisms can occur with torsional failure of thin disks, especially after such edge fracture surfaces have formed.³⁷ And so, while cookies may be nutritionally lacking, they present an enriching abundance of topics for scientific study.

We introduce Oreology (/ɔriːˈɒlədʒi/), from the Nabisco^® Oreo^® for “cookie” and the Greek rheo logia for “flow study,” as the study of the flow and fracture of sandwich cookies. We begin this study with a validation (Fig. 1) of previously reported²² suggestions for what causes the creme filling to separate predominantly onto one wafer or the other based on initial orientation within the product packaging. In Sec. II, we present equations to understand creme material deformation, torsional shear strain, and the resulting shear stress in counter-rotated parallel-plate mechanical tests. In Sec. III A, we test Oreos using a laboratory rheometer in a protocol also known as torsional gelometry when applied to cheeses.¹⁵ We monitor the shear stress–shear strain response to observe the characteristic failure response and report key mechanical properties of the creme filling. In Sec. III B, we investigate how the stress and creme distributions are affected by deformation rate (which mildly influences the resulting stress, indicating viscoelastic behavior), the amount of creme (which is not influential), and the flavor (which is not influential). In Sec. III C, we map out the spatial dependence of cookie failure orientation in relation to its position within a box. In Sec. III D, we investigate the observation that occasional boxes of cookies will contain cookies that predominantly all exhibit cohesive rather than adhesive failure of the creme. In Sec. III E, we explore time-dependent creep of the creme, which manifests as “delayed yielding,”³⁸ under which a constant applied stress causes Oreo creme to flow and break, not immediately but only after long times. In Sec. III F, we follow the degradation of chocolate wafer strength over time following milk imbibition, which causes significant structural loss in the first 30 s after contact.

When considering how to enable the broader accessibility of kitchen science, and rheometry overall, we must confront the fact that a laboratory rheometer is not widely accessible. For these reasons, “frugal science” and open-source measurement tools have taken off,³⁹ creating cost-effective research equipment such as origami microscopes;⁴⁰ wound string-powered centrifuges;⁴¹ LEGO-based microfluidic systems,⁴² microscopes,⁴³ and optical setups;⁴⁴ and 3D printed tools of many varieties including high-accuracy instrumentation for rheometry^12,45 and microfluidics.⁴⁶ And also, in Sec. III G, we present a 3D-printed torsion testing device designed for Oreos and similarly sized round objects. We conclude in Sec. IV with some ideas for future work and areas of study.

In laboratory rheometry, a sample fluid is typically placed between two coaxial parallel disks. The lower disk is held fixed while the upper disk is rotated at a constant rotation rate $Ω$⁠. This creates a wall-driven laminar (Couette) flow with internal tangential velocity ranging linearly in height from $v=0$ at the fixed lower disk to $v=rΩ$ at the upper disk, where $r$ is the radial position from the center of the disk [Figs. 2(a)–2(c)], and $z$ is the height above the stationary lower disk. The velocity field in the fluid is, thus,⁴⁷ 

For sandwich cookies, the analogy is apparent: the wafers are the parallel plates, and the creme is the fluid in between. When one wafer is fixed and the other is rotated, the central cylindrical disk of creme deforms until failure. Through analogy to this parallel plate setup, we calculate the material-level descriptors (shear stress, shear rate, and shear strain) for twisting Oreos, based on measured and applied quantities (torque and angular displacement). The shear rate $γ̇$ arising from the rotation rate $Ω$ with creme height $H$ will be

which varies from 0 at $r=0$ to a maximum value denoted by $γ̇R$ at $r=R$⁠, the creme radius:

The maximum shear strain $γR$ arising from the given angular displacement $θ$ will be⁴⁷ 

And the shear stress at the outer perimeter (⁠$r=R$⁠) corresponding to torque $M$ is

For simplicity of notation, we use $γ$ and $σ$ without subscripts to refer to these quantities, as we only consider the value at $r=R$ in the following measurements because we focus on failure, which initiates at the location of maximum stress/strain. In fact, due to this non-homogeneous strain, varying as a function of radius, additional corrections could be applied to recover more accurate measurements for large amplitude deformations.⁴⁸ For linear elastic materials at low strain (far below yielding), we expect the shear stress to be linearly related to the shear strain multiplied by the elastic shear modulus, $G$⁠, or

In our analysis, the height $H$ used in calculations refers to the height of the cream disk alone, even when a rheometer holds wafers as well because the wafers are significantly stiffer than the cream. We found the linear elastic shear modulus, $G$⁠, of the wafers to be at least $280$ kPa, while the modulus of cream filling is $≈$41 kPa, validating this approximation. The moduli were measured by small amplitude oscillatory shear tests on the rheometer.

When the cookie is rotated on the torsional laboratory rheometer (DHR-3; TA Instruments) at a constant speed (here, $Ω=0.1$ rad/s, or about 3.7 rotations per minute), the measured stress shows three mechanical regimes (Fig. 3). Initially, (1) a linear elastic response occurs, indicating a solid-like response of the creme to the applied strain. The stress–strain curve has an initial slope equal to the elastic shear modulus, $G$⁠. Next, (2) deviation from linearity indicates material plasticity and onset of irrecoverable yielding, and the stress maximum at the failure stress $σf$ and failure strain $γf$ marks material failure. Finally, (3) structural breakdown of cookie integrity propagates throughout the creme, typically due to creme delamination from wafer 2, which was affixed to the top rheometer plate for all our studies after identifying the native cookie-preferred orientation in each box by manual tests. Results are shown in Fig. 3 for several cookies each taken from two separate boxes purchased on separate days (red and black lines, respectively) with the average response (⁠$n≥4$⁠) for each shown in bold for clarity. Annotations indicate the failure stress $σf$⁠, failure strain $γf$⁠, and slope corresponding to the linear elastic modulus $G=σ/γ$ at low strains, which are also compared in Fig. 4 showing statistically different failure stress and linear elastic modulus, and statistically similar failure strain between cookies from the two boxes. The corresponding torque required to break open the Oreos in pure rotation is about 0.1 N·m. In comparison, the torque required to open a soda bottle cap is around 2 N·m, and a round doorknob may require 0.6 N·m to turn. Per NASA's human capabilities catalog, humans are able to apply twisting forces up to a range of 1 to 5 N·m using thumb and forefinger grip alone, or up to 15 N·m using full hand grip and forearm rotation.⁴⁹ By comparison, sandwich cookies are easy to open.

Within the realm of food science, the range of failure strains exhibited by the Oreo creme is on the brittle end, far below the failure strains of food gum additives and thickeners⁵⁰ and roughly equivalent to the failure strains observed in crumbly Romano or Old Amsterdam variety cheeses¹⁵ or foie gras.⁶ The failure stress of the Oreo creme is double that of a typical cream cheese⁵¹ or of peanut butter,⁵² and of the same magnitude as the failure stress of a soft Havarti or mozzarella cheese, but much less than typical cheeses, and so the creme is relatively soft. The failure stress and strain together place the Oreo creme within what has been previously identified as the “mushy” texture regime (as opposed to brittle, tough, or rubbery).¹⁵ 

Cookie connoisseurs may be prone to slowly or rapidly twisting their sandwich cookies (perhaps based on hunger or anxiety levels). To address the influence of these habits scientifically, in our laboratory rheometer, the upper wafers were rotated at a variety of controlled rates [$Ω=0.001 to 10$ rad/s, Figs. 5(a) and 5(b)]. We found that rotation rate influences the failure stress and strain by up to an order of magnitude, so that faster twisting results in higher yield stress and strain. This rate-dependence of the failure stress indicates a viscoelastic response in the creme, because a purely elastic or elasto-plastic failure would have a consistent failure stress at all applied rates. Rotation rate does not, however, influence the creme distribution—except at the highest testable rate, which tended to energetically free the creme from both wafers nearly simultaneously, and, in some cases, to also fracture the wafer [shown in the inset to Fig. 5(b)].

Next, the height of creme filling, or “stuf” level, was found to have only small influences on the failure stress and strain, as shown in Figs. 5(c) and 5(d). It was observed that double stuf cookies were filled to the largest outer radius, and so they required the most torque to separate, with some exceeding the torque limit of the laboratory rheometer (⁠$Mmax=200 μN·m$⁠) and causing greater uncertainty in these measurements. As has been previously observed, “double” and “more” stuf levels are in fact only about 1.9 and 2.7 times more stufed with creme by mass than the regular variety.⁵³ Finally, we analyzed the influence of creme and wafer flavors, comparing Golden, Dark Chocolate, and Team USA triple-stuf Olympic varieties, as shown in Fig. 5(e). Again, the required stress did not depend on creme/cookie flavor. Meanwhile, the wafer flavor influenced the likelihood of wafers to break. For instance, the Golden variety wafers fragmented during failure tests, causing creme to redistribute based on the wafer failure planes onto segments of both wafers. Two cookies are shown in Fig. 5(f) post-failure (but pre-consumption).

To understand cookie failure types within a full box of Oreos, all cookies were tested while keeping track of original location. In a single, freshly opened box of Family Size regular Oreos, which contains three rows of 16 cookies facing left–right in line with the label text, we found that cookies tend to have broken wafers near the box edges, creme staying on the right-most wafer in cookies on the left side of the box, and creme staying on the left-most wafer in cookies on the righthand side of the box (Fig. 6). Here, the terms “left” and “right” are oriented such that the package text is upright when the box is viewed from the front. The results show some randomness but average out to show consistent overall trends with several interesting features. The original spatial location of the Oreo in the box contributed a statistically significant impact (p < 0.05) on the creme distribution post-failure, with cookies on the left-most edge of the box statistically more likely to be fractured than cookies in other locations, cookies on the left-hand half of the box more likely to fracture with creme entirely on the right-most wafer, and cookies on the right-hand half of the box more likely to fracture with creme entirely on the left-most wafer (Fig. 6). Therefore, in addition to the observation that overall creme is most often oriented toward the right for a randomly selected cookie in this type of box, in a competing trend, the creme also falls most often on wafers facing inward into the box and away from the nearest (left or right) edge of the box. This may indicate that environmental impacts (ambient heat or mechanical perturbations) influence the cookies in individual boxes, having the greatest influence on cookies near the perimeter. As these influences are uncontrolled by individual consumers, this sensitivity may diminish the predictability of creme failure type and location between different boxes of cookies. No similar trends were observed comparing top-to-bottom distribution among rows rather than columns. How much this distribution changes between different boxes, cookie types, and season or other environmental impacts, and why, may be a topic for future study.

In almost all cases, we found cookies in newly obtained boxes to exhibit adhesive failure of the creme. However, after long times of being opened, or for the occasional newly opened box, nearly every cookie in the box split differently, leaving substantial creme on both halves. When this tendency toward cohesive failure was found to occur in a given box, cookies in these boxes tended to have creme filled between wafers to a greater radial extent without having greater total volume, as shown in Figs. 7(a) and 7(b). This suggested that creme had been allowed to flow after initial production, which would alter the creme itself and strengthen the creme–wafer bond enough to allow cohesive creme division. We could not pinpoint the specific reason for this difference, and systematic treatments of heat, humidity, staleness, mechanical agitation, compression, and preshearing only inconsistently recreated this effect. This change may have been due to variations in the manufacturing process, storage or shipping conditions with hotter temperatures, wafers growing stale, or other effects.

The stress profile measured for different varieties showed a single stress peak denoted $σf$ for adhesive failure [Fig. 7(c)]. Meanwhile, multiple peaks occur for cohesive failure due to the yield surface dynamically propagating with increasing strain. Because the maximum yield strain occurs at $r=R$⁠, inner sections of material at $r<R$ experience this yield strain at later time points when the top wafer is rotated at a constant angular velocity $Ω$⁠. In the case of cohesive creme separation, the creme failure surface was found to not lie parallel to the wafers but had a distinct “cup and cone” morphology, which would be expected for the failure of a brittle solid material, such as a metal, in torsion. This is because the failure plane for brittle materials is expected to be in the direction of maximum normal stress, which in torsional shear is along a 45° angle from the axis of rotation, thus forming a cone;⁵⁴ in contrast, failure for ductile materials would be expected for ductile materials in the direction of maximum shear stress, which in torsion is orthogonal to the axis of rotation. In addition, these cohesively failing Oreos followed the same orientation scheme observed previously; the “cup” side with slightly more creme tended to be on Wafer 1. Creme was distributed by cohesive failure evenly for all rotation rates, yielding 55% to 73% by mass on the more creme-laden wafer, with the creme division unaffected by the rotation rate, although a few cookies in this type of box would still exhibit adhesive creme failure.

While the fracture surfaces of ductile materials failing in torsion can be modulated by the applied torque, fatigue effects, and simultaneous tensile stresses^54–56 (i.e., combined loading, or peeling of an Oreo in addition to simply twisting), this was found to not be tunable for Oreos due to the apparent brittle nature of failure. To confirm this result, using a box of cohesively failing Oreos, we explored the effect of helical twisting profiles for failure. These paired a twist with a vertical push or pull with varying rate combinations of the two steps. No systematic trends were observed on the resulting creme distribution (data not shown). For other brands of cookies with more ductile or “creamier” fillings, it is theoretically possible to tune creme distribution by the kinematic profile of cookie twisting alone. It may also be achievable to tune creme distribution yet with Oreo creme at elevated temperatures when creme flows more easily, or through more careful heating and annealing of creme, which has been shown to influence whether colloidal soft matter will fail via brittle or ductile mechanisms.³² Although detailed exploration of such tuning is outside the scope of the current study, we performed exploratory experiments with lightly microwaved Oreos to confirm this possibility.

Yield stress fluids typically exhibit time-dependent behavior. This inclusion of time dependence on our Oreos may be considered a venture into a secondary new field, “horeology,” defined as the combination of horology (the study of time) with Oreology, defined previously. To assess how the Oreo creme distribution may change over time, we performed creep experiments with the laboratory rheometer, holding a constant torsional stress of low, medium, or high values (6.7, 15.5, and 19.1 kPa, respectively) relative to the expected yield stress of $≈$17 kPa, with results shown in Fig. 8. The creep compliance, or the quotient of the measured angular strain and applied stress, indicates the amount of material flow in response to the applied stress. When a high stress was applied, cookies yielded immediately shown by a sharply increasing compliance, scaling with time squared (thus being dominated by the acceleration of the upper rheometer fixture, rather than by material behavior). When a low stress was applied, cookies deformed very slightly (i.e., elastically) with some creep behavior emerging at longer times indicated by a change in the power-law slope. At moderate stress, this changing behavior was apparent as a dramatic shift in slope indicating delayed yielding. This phenomenon is also widely seen in creeping solids from mud to mustard^38,57 and has been implicated as a cause of landslides and avalanches due to sudden catastrophic material flow even under constant applied stress. Further, at elevated temperatures, creep is expected to increase, and this phenomenon may partially explain how creme may spread over time as in Fig. 7 while wafers remain intact.

Once a cookie has been separated, one is typically left with a wafer with creme and a bare wafer, which both may be dipped in milk to enhance final consumption. A wafer in contact with milk will imbibe through wicking of the liquid into its porous structure, which quickly weakens.⁵⁸ The laboratory rheometer can sensitively probe the mechanical properties of such material mixtures as they change over time. We affixed a single wafer to our laboratory rheometer using a thin layer of glue to monitor this process. We measured the modulus of the glue to be three times larger than that of the wafer, and the glue layer was under one tenth of the height in total, and so deformation in the glue layer can thus be safely neglected. Low-fat (1%) milk was added by pouring around the perimeter of the wafer, and the linear elastic shear modulus was monitored using a small amplitude oscillatory shear test at a small strain (⁠$≈$0.1%). Results show a slight increase in the linear elastic modulus due to swelling, followed by a decrease from an initial value of 560 kPa to a final value of 1 kPa with most change occurring over the first 30 s (Fig. 9). We connected this mechanical decline with manual experiments in which we dipped a single wafer into 1% milk for 5 s to ensure sufficient takeup of liquid⁵⁸ and held it by hand suspended from one point on the rim. In this manual test, the wafer was observed to break under its own weight after an average of 30.2 s (⁠$n=8$⁠), shown by a dashed vertical line in Fig. 9, and corresponding to a loss of mechanics of about 95%. Luckily, in a full cookie, sufficient creme remains to hold cookies intact for longer times (an average of 63 s, $n=6$⁠, under the same testing protocol), as the (nonporous) creme is not similarly weakened mechanically by the milk. However, the orientation and style of manually dipping a cookie or biscuit into liquid will influence the rate of imbibition and subsequent mechanical collapse, as has been reported before.¹⁶ Although this is beyond the scope of the current study, there is likely a “sweet spot” between immersion time to load the cookie with milk and consumption time before loss of integrity. The use of higher-fat milk is also expected to lead to a slower uptake of the liquid due to the milk solids clogging pores in the wafer, as has been shown previously for cereal in milk,⁵⁹ which may make this “sweet spot” easier to find. How the macroscale textured ridges on wafers influence fluid imbibition would also be an interesting area of future study.

Explorations of cookie mechanics varying time, rate, and stress using the laboratory rheometer provide a taste of the potential applications of rheology to quantify and thereby understand flow and fracture of sandwich cookies. To enable further studies beyond the laboratory, we introduce the Oreometer, which is a 3D-printed torsion testing device designed for Oreos and similarly dimensioned round objects (Figs. 10 and 11, Multimedia view). The device requires no power or electronics and has a material cost of $6 (including coins and rubber bands), enabling widespread use.

In the Oreometer, cookies are mounted into two rubber-band powered clamps held on a larger fixture. The rubber band pattern adjusts the holding torque on the wafers, as the clamping force tightens when the band is strained by winding around more posts [Fig. 10(c)]. For brittle cookies with variable sizes, it was important to find the rubber band configuration that held the wafers securely (i.e., to apply enough torque to complete the measurement) without fracturing wafers due to high clamping forces. Next, pennies (quantity $N$⁠) are loaded into one of the two symmetric “penny castles,” applying torque in precise increments [Figs. 10(a4), 10(a5), 10(c4), and 10(c5)]. Staggered castle windows, or balistraria, in the penny castles have heights equal to five pennies to expedite counting in groups of five. The friction torque in the Oreometer opposing twist of the cookie is overcome by less than one penny's weight, $mcoing$⁠. Torque comes from the applied force $F$ along the lever arm between the penny castles and center of rotation, $L=90 mm$⁠, and the angle $θ$ between the lever arm and the force of gravity. The applied torque $M$ is then

Assuming the Oreometer is standing upright without twisting so that $θ≈90°$ as in Fig. 10(a4), the applied torsional stress to the cookie inside is then

The stress calculation does not depend on creme height (which influences only the material strain field, not the stress). A video illustration of this device is included in Fig. 11 (Multimedia view) along with the supplementary material containing design files in .stl and .step formats, and a step-by-step tutorial to use the Oreometer for further investigations.

Cookies from both boxes used in Fig. 3 were tested to failure using the Oreometer, and data were compared to those from the laboratory rheometer. We found that the failure stress measured by the Oreometer is not statistically significantly different (p > 0.05) from the set of rheometer data [Fig. 10(b)]. Creme distribution results from cookies that were tested to failure on the two devices were also comparable. It was further observed that cookies left in the Oreometer unyielded but with some torque applied may spontaneously yield after some time without additional coins being added. This is due to the same delayed yielding effect reported in Fig. 8.

Through a series of experiments with a laboratory rheometer used to hold whole Oreo cookies, we determined that creme distribution upon cookie separation by torsional rotation is not a function of rate of rotation, creme “stuf” (i.e., filling height $H$⁠) level, or flavor, but was mostly determined by the preexisting level of adhesion between the cookie creme and each wafer. In most cases, creme delaminated from the wafer with a preferential orientation with respect to the package within any one box, allowing prediction of failure direction with 80% accuracy. Despite the consistent failure mode, there was some amount of cookie-to-cookie and box-to-box variation in failure stress and strain. Apparent reflow of creme due to unknown causes had the most significant effect in altering failure type, allowing for improved creme–wafer bonds and subsequent cohesive failure of the creme, splitting nearly evenly between the two wafers. Failure mechanics further allowed us to classify the creme texture as “mushy.”

Moreover, questions remain to be explored if we are to fully understand cookie flow phenomena, and for this purpose, we have introduced a home-use Oreometer. A topic of particular interest may be what influences the change from adhesive to cohesive failure. If this is due to altering of the creme structure, then one may wish to explore whether it can be affected by compressing cookies before twisting, humidity, temperature, addition of a flow agent, or fracturing wafers before twisting. Other tests could be done with temperature control by using the Oreometer on cookies quickly withdrawn from the fridge, microwaved slightly, or dipped immediately beforehand in milk, where the device can be placed close to the kitchen appliances to minimize transfer time. The mode of failure of cookies may also be investigated manually by peel tests applied at the same time as torsion to study mixed-mode failure. If cookie manufacturers would like to influence creme distribution themselves, providing wafers with through-holes or texture on inner surfaces should promote creme–wafer adhesion onto both halves. Future studies may examine other sandwich-variety cookies, such as other brands of sandwich cookie, custard creams, macarons, and ice cream sandwiches—wherein temperature would be expected to have greater influence—as well as Nutter Butters, peanut butter between crackers, and other savory snacks. Planar shear tests, rather than torsional shear, may probe whether the fruit paste inside Newtons (previously known as Fig Newtons) indeed have the implied Newtonian flow behavior, or whether they are not only sometimes non-fig-containing, but are also rheologically non-Newtonian [and hence may become known as non-Newton(ian)s]. Our results and methods of investigation may also have widespread application in understanding other torsional events in the kitchen, from braided breads and mixing dough to ideal opening kinematics of stubborn jam jar lids. Finally, the development of an Oreology subdivision within the Society of Rheology might be expected as this field of study evolves and expands.

Rheological behavior was measured using a stress-controlled shear rheometer (DHR-3, TA Instruments) using a parallel plate fixture (40 mm diameter; TA Instruments) at a controlled temperature of 25 °C. The cookies and creme were initialized with a 0.5 N compressive load after loading in the rheometer to ensure a consistent start point, at which point the vertical gap was fixed and the experiment was performed allowing the normal force to vary. Oreos^® (Nabisco) were obtained from the local grocery store, and the wafers were firmly adhered to the plates using thermoplastic adhesive from an “ultra-low temperature” hot glue gun (AdTech, #05690, temperature <100 °C). This glue was measured by small amplitude oscillatory shear to have an elastic shear modulus of 1.4 MPa. Cookies with broken wafers were typically not used. Reported modulus, yield stress, and yield strain were extracted from the stress–strain curves, and statistics were performed on these values using an unpaired two-tailed t-test. P-values of less than 0.05 were considered significant.

Designs were made in Autodesk Fusion 360 and printed out of PLA using a Creality Ender 3. Parts were sliced with Cura using a 0.3 mm layer height, 20% infill, 210 °C extruder temperature, and 55 °C bed temperature, and printed with supports for the base halves, Oreo clamps, and penny castles.

See the supplementary material for 3D printing design files and a user-friendly tutorial on getting started with the Oreometer.

C.E.O. was supported by the United States Department of Defense (DoD) through the National Defense Science and Engineering Graduate Fellowship (NDSEG) Program (OMB 0701–0154) and the MIT MathWorks Engineering Fellowship. M.R.F. was supported by funding from MIT's Undergraduate Research Opportunities Program. We further gratefully acknowledge delightful conversations with Cody Moose (who coined both the paper title and the name of our in-lab cookie sample repository, the “drawereo”), Jake Song (who first introduced adhesion terminology into our story), and Rubens Fernandes, Randy Ewoldt, and others at the Society of Rheology Annual Meeting in 2021 for insightful commentary about what is a fluid and what is not.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.