mixtures may be prepared by volumetric dilution of pure carbon monoxide with nitrogen or helium, which are free of carbon monoxide. If the volumes used in dilution are accurately known, the concentration may be calculated. A known small volume of pure carbon monoxide is placed in an evacuated tank of known volume, and the tank is then filled with the dilutor gas. The pressures of the carbon monoxide volume and the final mixture must be known (or else known to be equal). Smaller samples may be prepared in plastic bags by injecting pure carbon monoxide from a gas syringe into a stream of dilutor gas metered accurately into the bag with a flow device such as a wet test meter. Because cylinder nitrogen commonly has small amounts of carbon monoxide present in it, it must not be assumed to be pure without verification. Pure helium is reliably free of carbon monoxide. A useful procedure is to zero the detector using helium. After that, one can measure the CO content of the available cylinder nitrogen and can make the required correction. This permits using the less expensive nitrogen for most calibrations. A reference gas mixture whose carbon monoxide content is not accurately known can also be analyzed gravimetrically or volumetrically by various methods, but these techniques are usually less convenient and often require large amounts of gas.When very accurate low-level measurements are needed, or when background gases that have the potential to interfere with the measurement are present, gas filter correlation (GFC) is used (Figure 8.10c). In these designs the measuring and reference filters are replaced by gas-filled cuvettes. The reference cuvette is filled with CO and the measuring cuvette usually with nitrogen. In addition to being unaffected by the presence of background gases, both the accuracy and the response time of these instruments are better than those using filters. If DFC is used in combination with single-beam, dual-wavelength technology, it is virtually immune to obstruction of the optics.