After induction of anesthesia, they noticed a "cleft" ( a sudden dip in CO2) during expiratory plateau phase on capnogram. Additional vecuronium was administered. There was no obvious inspiratory efforts by the patient nor was the negative deflection of the airway pressure gauge typical of inspiratory effort. The ETCO2was observed to be increased from control value of 33 mm Hg. There was also a decrease in peak airway pressure as well as a decrease in tidal volume. A leak in the breathing circuit was suspected and found between the capnograph main stream sensor assembly and endotracheal tube. Reconnection of the transducer eliminated the leak and restored the capnogram to normal in the next exhalation.

The authors provide the following hypothesis.

In the presence of the disconnection, the inspiratory capnogram is unchanged. During expiration, the initial high expiratory flow rate will be sufficient to close the inspiratory valve and allow exhaled gas to flow through the CO2 analyzer giving rise to the normal looking expiratory plateau. However, a portion of expiratory gases exit through the leak. This reduces the measure expired tidal volume. As the expiratory flow rate decreases, leak flow will cause pressure to decrease in the circuit, allowing expiratory valve to close prematurely and facilitate flow of inspiratory gases.  The flow of fresh gases flow from the inspiratory limb reduces the measured CO2  at the analyzer. As the fresh gas flow causes pressure to rebuild in the circuit, the expiratory valve will reopen, allowing the further movement of alveolar gases pass through the CO2 analyzer.

Reference:
Tripathi M, Tripathi, M. Partial disconnection at the main stream CO2 tansducer mimics

'curare cleft" capnogram. Anesthesiology 1998;88:1117-9.