Why there's always a test better than TPA

Texture Profile Analysis is one of the most recognised tests in food texture measurement - and for good reason. In a single run, it produces a suite of parameters: hardness, cohesiveness, springiness, resilience, adhesiveness, gumminess and chewiness. Set it up once, press go, and the software does the rest. It's fast, it's familiar, and it feels comprehensive.

But that ease of use has become something of a trap. Across laboratories worldwide, TPA has quietly evolved from a carefully applied scientific tool into a default setting - selected not because it's the right test, but because it requires the least thought. And when a test is chosen for convenience rather than suitability, the data it produces can be misleading at best and meaningless at worst.

This post isn't an attack on TPA. It's an invitation to think more precisely - because for almost every sample you're working with, there is a single, more focused test that will tell you exactly what you need to know, without the noise.

What TPA was designed to do

It's worth remembering what TPA actually is: a double compression test, using a probe larger than the sample, that simulates the mechanical action of a first and second bite. The parameters derived from the curve were developed and validated to correlate with sensory evaluation of foods that behave like they do in the mouth - being bitten, deformed significantly, potentially fractured, and chewed.

The method was born in the 1960s. Its creators were rigorous. Alina Szczesniak - one of TPA's founding scientists - later wrote a letter to the Journal of Texture Studies expressing deep concern about how the test was being applied. She described seeing penetrating needles used instead of compression plates, chewiness calculated for hard candy (which is sucked, not chewed), and springiness reported for materials that had been crushed during the test. Her conclusion was that misuse was widespread and that standardisation was urgently needed.

If one of the method's own founders felt compelled to issue that warning, it's worth pausing to ask: how many laboratories today are making the same mistakes?

The comfort of collecting everything

The appeal of TPA is understandable. In one test, you appear to get seven parameters. There's a sense of thoroughness to it - a feeling that you've measured your sample comprehensively. But this is precisely the problem.

Not all TPA parameters are relevant to every sample. When you test chocolate, springiness values are unlikely to be repeatable or meaningful - chocolate is not a springy material. When you test bread, adhesiveness is not an important textural characteristic and should not be reported. When you test a hard candy, gumminess and chewiness are derived from a material that in reality is never chewed at all.

The TPA macro will always produce a value for every parameter it is designed to find. It has no way of knowing whether those parameters are appropriate for your sample. That judgement belongs to you. And if you accept all seven numbers without first asking which ones actually matter for your product, you are not conducting rigorous science - you are generating noise that looks like data.

The hidden limitations of compression-only testing

There's a deeper structural issue with TPA that goes beyond parameter relevance: it is, by design, a compression test. A flat probe presses down on a sample. Twice. That's it.

The texture of real products, however, is rarely captured by compression alone. Consider what actually happens when a product is evaluated - by a consumer, a panel, or a quality team:

A sausage is bitten and sheared. A biscuit snaps. A yogurt is spooned and spreads. A tablet is pressed out of a blister pack. A gel is squeezed from a tube. A confectionery coating cracks. A bread roll is pulled apart. A film is stretched.

None of these actions are pure compression. None of them are well-represented by pressing a flat probe vertically downwards onto a cut cube of product. Using TPA for these materials means forcing a complex, multi-dimensional textural experience through a single mechanical lens - and then wondering why the data doesn't correlate well with sensory scores or quality outcomes.

The Texture Analyser is capable of an enormous range of test types: tension and extensibility, cutting and shearing, bending and snapping, puncture and penetration, extrusion, adhesion, spreading, and more. Each of these, applied with the right attachment, gives you a measurement that is directly relevant to how your product actually behaves. TPA gives you a compression measurement dressed up in sensory language.

The parameter settings problem

Even when TPA is the appropriate choice, the test is far more sensitive to parameters than most operators appreciate - and inconsistency in those parameters makes results incomparable across runs, batches or laboratories.

The extent of compression matters profoundly.

TPA was designed to reach deformation levels sufficient to break the sample - to genuinely simulate mastication, which is a highly destructive process. Compressing a sample to only 10-30% and reporting cohesiveness, springiness and chewiness as though they were derived from a valid TPA is a common error. At low deformation, those parameters carry none of the meaning the method was designed to confer.

Test speed affects results significantly.

A slower speed allows more stress relaxation in the sample, reducing the measured force. The pre-test speed affects trigger accuracy. The time between the two compression cycles influences every time-dependent parameter - springiness, cohesiveness, gumminess and chewiness are all sensitive to the inter-bite interval. If you change the speed or timing between tests, your results are not comparable.

Sample dimensions matter too.

Hardness as recorded by TPA is not an absolute material property - it changes with sample height, contact area, and the percentage compression applied. Two samples of the same material but different size will produce different hardness values. Comparing TPA data across studies that used different sample geometries is scientifically unsound.

These are not obscure technical points. They are the fundamental conditions under which TPA is valid. Without controlling them rigorously, the test produces numbers that feel precise but mean very little.

There is always a better test

Here is the most useful thing to take from all of this: for every sample you're testing, there is a single property that matters most - and there is a test that measures that property directly, cleanly and without irrelevant parameters clouding the picture.

Are you trying to understand how firm a gel is? A single compression to a defined deformation will tell you that, with none of the complications of the second compression cycle.

Are you interested in whether your product fractures or snaps? A 3 Point Bend test gives you a clean fracture force and snap profile that a compression test simply cannot capture.

Are you measuring the spreadability of a cream, paste or soft cheese? An extrusion or back-extrusion test gives you a direct measure of flow and spreadability under controlled conditions.

Are you evaluating how well a coating or film stretches before breaking? A tensile test with grips gives you extensibility and tensile strength - data that is directly meaningful for your application.

Are you assessing the stickiness of a confectionery product or adhesive? A dedicated adhesion test - probe tack, peel, or adhesive bond - gives you the energy and force of separation in a way that TPA's adhesiveness parameter, measured during withdrawal from a compressed sample, simply cannot replicate with the same fidelity.

Are you measuring the crunchiness of a snack? A puncture or shear test, potentially combined with acoustic detection, will capture the fracture event and the sound energy that your consumers actually experience.

In every case, the targeted test gives you a measurement that is:

• Directly linked to the physical mechanism you care about
• Free of parameters that don't apply to your sample
• Not constrained by the need for a flat compression probe and a bite-sized cube
• Easier to validate against sensory data, because the test mode matches the sensory experience

A more thoughtful approach

The next time you set up a TPA test, ask yourself one question first: what is the single most important textural property of this sample?

If the answer is hardness, run a single compression. If it's extensibility, run a tension test. If it's snap, run a three-point bend. If it is genuinely a combination of hardness, cohesiveness and springiness that you need to compare across a range of similar products - and those products really do behave like foods being masticated - then TPA may well be the right choice, applied carefully and with full awareness of the conditions that make it valid (and likely with some of the measured irrelevant parameters removed).

But if you're running TPA because it's quick, because the macro is already loaded, or because it produces seven numbers at once - consider what you're actually measuring. Are those numbers telling you something true and useful about your product? Or are they the texture equivalent of asking a comprehensive questionnaire when one well-chosen question would give you everything you need?

The Texture Analyser is one of the most versatile instruments in any laboratory. It can answer almost any question about the mechanical behaviour of a food, material or product. TPA uses only a fraction of that versatility. There is almost always a test that will answer your specific question more directly, more cleanly, and with more confidence.

The best measurement is not the one that produces the most parameters. It's the one that tells you exactly what you need to know.

Other technical points to note when using TPA

On the fundamental origin and intent of TPA:

• TPA was originally developed to mimic the action of the human jaw - two bites on a bite-sized piece of food. This means it was conceived specifically for foods that are eaten by chewing. Applying it to materials that are never placed in a mouth (pharmaceutical tablets, cosmetic creams, packaging films, adhesives, polymers) stretches the method far beyond its conceptual foundation. The sensory language of the parameters - hardness, chewiness, gumminess - becomes meaningless when the material is never experienced sensorially in that way.

On the compression probe limitation specifically:

• TPA was originally developed with the specific intention of using a probe larger than the sample for true uniaxial compression - yet many published papers have used penetrating/puncture probes and still called it TPA. This is a fundamental methodological error that invalidates the entire test name and its parameter meanings. A puncture causes completely different structural damage to compression, and penetrating the same spot twice produces meaningless data.

On gumminess vs chewiness - a commonly missed error:

• These two parameters are mutually exclusive by design, yet are routinely reported together for the same sample. Chewiness applies to solid foods; gumminess applies to semi-solid foods. Reporting both for the same sample was explicitly flagged as a misunderstanding by TPA's own founders - yet it persists widely.

On the post-test speed being overlooked:

• The post-test speed must match the test speed for cohesiveness to be calculated correctly - a detail that is frequently overlooked and silently corrupts results.

On the auto-acceptance problem:

• The software always produces a value for every parameter. There is no flag or warning when a parameter is inapplicable. Springiness for chocolate, adhesiveness for bread - the macro reports them anyway. The operator must supply the scientific judgement the software cannot. 
• As data analysis becomes increasingly automated, there is a growing risk that TPA parameters are fed directly into reports, dashboards and decision systems without any human review of their appropriateness. The risk of a parameter being flagged as out-of-specification when it was never meaningful for that product in the first place is a real quality management concern.

On inter-laboratory comparability:

• Because TPA results are so sensitive to sample geometry, compression percentage, speed and timing, TPA data from one lab is rarely directly comparable to TPA data from another - even on the same product. This is a serious limitation for any work intended to reference published benchmarks.

On derived parameters being mathematically dependent, not independently measured:

• Gumminess and chewiness are not measured at all - they are calculated from other parameters (hardness × cohesiveness, and hardness × cohesiveness × springiness). This means any error or irrelevance in hardness or cohesiveness is multiplied and compounded into those derived values. Researchers who report gumminess and chewiness as though they are independent measurements are misrepresenting the nature of the data.

On the trigger force and contact point being critical but often wrong:

• If the pre-test speed is too fast, the probe overshoots the surface and enters the sample before data collection begins. This means the initial contact - the very start of the compression curve - is missed, and hardness and fracturability values are compromised from the outset. This is a silent error that produces no warning message.

On sample heterogeneity being invisible to TPA:

• TPA compresses a single cube of product. If the product is heterogeneous - a filled biscuit, a layered confection, a meat product with fat and lean - the probe encounters different structures at different points and the resulting curve is a confused composite. A more targeted test (puncture to a specific depth, shear through a specific layer) can isolate the structural element you actually care about.

On the hold time between compressions being rarely standardised:

• The time between the first and second compression is a variable that profoundly affects springiness and cohesiveness in viscoelastic materials, yet it is rarely reported in published work and rarely standardised within a laboratory. Two operators running what they believe to be the same method may produce systematically different results simply because this value differs.

On TPA being unable to capture dynamic or rate-dependent behaviour:

• Many food and material properties are rate-dependent - they behave very differently at slow versus fast deformation rates. A single TPA test at one speed gives you no information about this viscoelastic behaviour. Creep tests, stress relaxation tests, or tests at multiple speeds would reveal far more about how the material will perform under the range of conditions it encounters in real use.

On the inability to test irregular or asymmetric samples:

• TPA requires a consistently sized, regularly shaped cube or cylinder. Products that cannot be cut to uniform dimensions - whole fruits, irregularly shaped snacks, products with natural variation in size - either cannot be tested by TPA at all, or require so much sample preparation that the prepared sample no longer represents the product as consumed.

On the compression platen missing surface and skin effects:

• A flat compression probe averaging force across the entire top surface of the sample will mask localised effects - the crispness of a crust, the skin of a sausage, the coating of a tablet. A puncture or penetration test, by contrast, isolates exactly the surface layer behaviour that determines first-bite consumer perception.

On TPA not capturing fracture mechanics properly:

• Fracturability is listed as a TPA parameter, but it only registers if there is a distinct peak before the maximum force peak in the first compression. Many brittle or snapping products fracture in a way that is far better captured by a three-point bend, a snap test, or a Kramer shear test - all of which give clean, unambiguous fracture force and energy measurements without the structural confusion of compressing an already-fractured sample for a second time.

On adhesiveness being one of the weakest TPA parameters:

• TPA adhesiveness is measured during probe withdrawal after compression - meaning the surface has already been significantly deformed and damaged. True adhesion testing (probe tack, peel, or controlled adhesion/cohesion tests) measures adhesion on an undamaged surface under controlled approach and separation conditions. For any product where adhesion or stickiness is a primary quality parameter, TPA adhesiveness is a very poor proxy.

On the lack of directional sensitivity:

• Compression is a single-axis, single-direction test. Many products have directional or anisotropic texture - meat fibres, layered pastry, extruded snacks, woven textiles. Compressing these samples vertically tells you nothing about their resistance to shear, their layered delamination behaviour, or their fibre-direction mechanical properties. Targeted tests with appropriate geometries capture these distinctions; TPA cannot.

On TPA being unsuitable as a QC tool in many settings:

• Because TPA requires careful sample preparation, consistent dimensions, and tightly controlled test parameters, it is poorly suited to high-throughput quality control environments where speed and simplicity are essential. A single-parameter test - a puncture force, a snap force, an extrusion force - is faster, more robust, less sensitive to operator variation, and often more directly linked to the specific quality attribute being controlled.