More than 33 m of 2.5 Ma sediment from Searles Lake, California was studied in order to construct a record of secular variation (SV) across the Gauss/Matuyama (G/M) normal-to-reverse polarity transition. The behavior of the field preceding and following the reversal is considered here, while in a companion paper <Glen et al., this issue> the details of the transition are discussed. The record encompasses an interval of roughly 183,000 years beginning 50 kyr (9 m) before and extending more than 128 kyr (23 m) beyond the transition, while the main phase of the transition lasts for nearly 5 kyr (1 m). Because the core was rotary drilled, and declinations lost, SV was characterized by the inclination and its angular dispersion. Inclination-only statistics reveal that (1) the record displays overall higher than expected values of angular dispersion (normal S~20¿; reverse S~19¿; expected S~15.5¿), suggesting that the field proximal to transitions may be more noisy than the distal field. In addition, normal data from immediately before the transition display higher S than reverse data immediately following it, implying that the postransitional field is more stable than the pretransitional field. One of the most prominent features of this record is an excursion of the field occurring roughly 4 kyr prior to the onset of the reversal. A record of the G/M transition from Chinese loess (R. Zhu et al., submitted manuscript, 1999) displays a similar event (also occurring roughly 4 kyr before the transition). This and the fact that the event is associated with anomalously low intensities suggest that the disturbance may be global in nature. The fact that comparable features are associated with other transitions <Hartl and Tauxe, 1996; Clement, 1992> intimates that the field may commonly show signs of early instability. This precursory event is actually one of a sequence of oscillations (in inclination and intensity) preceding the transition. That these fluctuations occur at roughly 4 kyr intervals leading up to the reversal (which also appears at this same interval) strongly suggests that an oscillatory disturbance in the core, active over at least 15 kyr prior to the transition, had eventually triggered the reversal. In addition, that these waveforms are absent from the postransitional record suggests the reversal process actively rejuvenates and stabilizes the field. ¿ 1999 American Geophysical Union