Best practices

Validating a liquid class: gravimetric calibration, step by step

A liquid class is a hypothesis until you weigh it. How to verify accuracy and precision gravimetrically, adjust with a calibration curve, and re-check over time.

Every liquid class starts life as a guess: a set of parameters that ought to deliver the right volume. Whether it actually does is an empirical question, and the most direct way to answer it is to weigh what comes out. Gravimetric validation turns a plausible class into a trusted one, and skipping it is how systematic short-fills hide in a workflow for months.

Why weighing beats eyeballing

A dispense can look perfect and still be five percent low. The eye cannot resolve the small, consistent errors that matter most, because they are the ones that bias every replicate in the same direction. A balance can. Weigh the dispensed liquid, divide by its density at the measured temperature, and you have the delivered volume to a precision the eye never reaches.

A three-step workflow

  1. Screen a starting class. Begin from a predefined class for a similar liquid and volume range and dispense a set of replicates across the volumes you care about. This tells you how far off you are before you change anything.
  2. Adjust against a calibration curve. Weigh the results, compare delivered volume to target across the range, and fit the relationship. Use it to correct the systematic offset, typically through the class calibration or correction curve rather than by guessing at flow rates.
  3. Confirm. Run a fresh set of replicates with the adjusted class and verify that both accuracy and precision now meet your acceptance criteria. If they do not, iterate; if they do, record the result.

Read accuracy and precision separately

The weights give you two numbers, and they call for different fixes. Accuracy is how close the mean delivered volume is to the target, and a consistent offset is corrected with the calibration curve or over-aspiration. Precision is the spread between replicates, usually reported as a coefficient of variation, and it is improved by the parameters that stabilize the transfer: flow rate, delays, air gaps, and level tracking. A class can be precise but biased, or accurate on average but noisy, and only separating the two tells you which knob to turn.

Control the things that quietly move the result

  • Evaporation: small volumes lose mass to the air fast, so use an evaporation trap or a lid and work quickly, especially with volatile liquids.
  • Temperature: density and viscosity both shift with temperature, so measure the liquid temperature and use it in the volume calculation.
  • Balance resolution: the balance has to resolve the smallest volume you dispense, or the low end of your range is just noise.
  • Replicate count: a handful of dispenses cannot characterize precision. Weigh enough to trust the coefficient of variation.

Validation is not a one-time event

A class that passed six months ago may not pass today. Tips change lots, ambient conditions drift, and instruments age. Periodic re-verification, and a re-check after any hardware service, is what keeps a validated class honest. In a regulated environment it is also the evidence that your automated results are defensible.

A liquid class is only validated for the instrument, tips, volumes, and liquid you tested. Record those conditions with the numbers, or the validation means little to the next person.
Piptera

Notes on pipetting calibration, liquid classes, and building an open, vendor-neutral catalog for every liquid handler.

© 2026 Piptera. Built for labs.