Sampling a finished blend

Sampling is where uniformity stops being an intention and becomes a measurement, and the way most small manufacturers do it cannot detect the failure they are worried about.

One sample proves nothing

A single sample from the top of a blender tells you the average composition of that spoonful. It cannot tell you whether the batch is uniform, because uniformity is by definition a comparison between locations.

If the whole point is that every serving matches, the test has to be between servings.

Where to sample from

Multiple points spanning the batch: top, middle and bottom of the blender, and ideally from different horizontal positions too. Sample after discharge as well, because discharge is itself a segregation event.

Then again from the start, middle and end of the filling run, because a hopper segregates as it empties and the last tubs can differ from the first.

A thief or sampling probe takes material from depth rather than from the surface, which is what you need.

What to do with the samples

Compare them. For a flavored product, taste them side by side, which is fast, free and detects segregation of the small components better than most instruments.

For anything where potency matters, assay them. The variation between those samples is your blend uniformity, and it is the number that says whether the panel is true per serving or only on average.

Reserve samples are separate

21 CFR 111 expects reserve samples of finished batches held for a defined period. That is a different purpose: not to prove uniformity now but to answer a question later.

Keep them sealed, labeled with the batch code, stored as the product should be stored, for at least the declared shelf life. They are the only way to examine what you actually shipped rather than what you recorded.

References