Two theories account for when and how people realize the sequence rules in implicit learning: the continuous strengthening of the memory characterization and the unexpected-event hypothesis. The latter theory supposes that the generation of the explicit knowledge of an accidentally experienced rule is due to a controlled search as well as the regular task processing. This search is motivated by unexpected events that occur during task processing. For instance, while performing an accidental sequence learning task smoothly subjects may unexpectedly experience an unmatched feeling, and while examining causes of the unexpected experience they find more regular patterns in the task. Based on this theory, this study set transfer chunk to motivate the search for accidents. We adjusted the number and position of transfer chunks to investigate how to enhance efficiency in implicit sequence learning. We adopted the serial reaction time (SRT) tasks and increased the number of transfer chunks and changed their position, assuming that the more transfer chunks there are, the more amount of implicit knowledge and the higher level of consciousness subjects will get and that the earlier transfer chunks appear, the more amount of implicit knowledge and the higher level of consciousness they will get. A total of 86 undergraduates took part in the experiment of 10 training chunks with a four-choice SRT task. Each chunk consisted of 96 trials, and so 960 trials in total. On each trial, subjects responded to the location of the target as quickly and accurately as possible by pressing the corresponding key. In this study, subjects in every condition had to finish the training chunks with the SOC1 sequence and the transfer chunk with the SOC2 sequence. In addition, conditions 1 and 2 both had only one transfer chunk, with an earlier position of the transfer chunk in the sequence in condition 1 than condition 2. Conditions 3 and 4 both had two transfer chunks, with the first transfer chunk appearing earlier in condition 3 than condition 4. Subjects had to finish the learning phase first, then make a verbal report, and finally finish Include and Exclude tasks. According to the verbal reports, we selected the data which were not completely explicit, and then analyzed their amount of implicit learning and level of consciousness. The results show: (1) a significant effect of number, that is, two transfer chunks can better promote implicit sequence learning, which proves that novel stimuli can play a role of "accidents" only with a certain quality of the characterization; (2) as to the position effect, two transfer chunks with the earlier position of the first transfer chunk can generate more explicit consciousness, which means that the first novel stimulus has to appear in the phase of primary quality characterization, and that the comparison of the first novel stimulus with the original sequence and echo with the second novel stimulus can promote the level of consciousness of the original sequence.