It is generally assumed that native deoxyribonucleic acid (DNA) is a right-handed double helix. A reasonable deduction is that during replication, the two parental strands have to unwind very quickly. However, this surmised quick unwinding is problematic and has never been proven experimentally. It is hypothesized that the two strands of DNA are winding with each other ambidextrously rather than plectonemically. The successful assembling and disassembling of a zero linking number topoisomer supports this hypothesis. It was further proven by quick separation of singly nicked DNA. The new DNA model was also verified by the "figure 8" structure, which is the annealing product of two single-stranded circular DNA with a 2 kb complementary insert in opposite directions. These experimental results are hard to be explained by the traditional Watson-Crick model. The significance of this finding in the understanding of DNA replication is briefly discussed.