How repeated readings and multi-item sampling answer different quality questions.
The two approaches measure different sources of variation
Repeatedly measuring one item mainly shows how much readings vary when the item is intended to remain the same. Depending on the procedure, that variation may reflect the instrument, operator, fixture positioning, environmental conditions, or short-term changes in the measurement process. It does not reveal how much products differ from one another because only one product is represented.
Measuring multiple items addresses product-to-product variation within the sampled group. Each item may also be measured repeatedly, but those extra readings do not increase the number of products sampled. Ten readings from one item remain evidence about one item, not ten independently selected items.
The records must preserve both counts
A clear report states the number of distinct products and the number of measurements taken from each product. Sample identifiers should connect every reading to the correct item. The procedure should also say whether an item was removed and repositioned between readings, whether operators changed, and how the final value for each item was calculated. Without this structure, instrument repeatability and product variation can become mixed together.
NIST Technical Note 1297 supports the general practice of evaluating a result together with measurement uncertainty. NIST's discussion of lot acceptance sampling adds a separate distinction: sampling from a lot is intended to support a decision about that lot, not to estimate every aspect of its quality. Thus, neither repeated readings nor a limited multi-item sample should be made to support a broader conclusion than its design allows.
The practical conclusion
Use repeated measurements of one item to study the consistency of the measurement process under defined conditions. Use multiple independently identified items when the question concerns differences among products or a sampled batch. A SEWARD-specific conclusion would additionally require batch-linked sample identifiers, individual readings, and the applicable repeatability plan.