A key issue about the Arctic Oscillation (AO) is whether variations over the Arctic (N), Atlantic (A), and Pacific (P) regions are truly coherent (true AO), or two independent seesaws cause an apparent seesaw over the three regions (apparent AO). We make data analyses as to which is real. Analyses excluding N show that variations in A and P are independent. With including N in the analysis, the AO and a negative correlation mode between A and P simultaneously emerge. Next, two time series, which vary with the AO pattern in A and P, are produced. The correlation between the two is small, but the sum of the two and the time coefficient of the AO highly correlate. All these results lead to the conclusion that the observed AO is almost apparent. In practice, two almost-independent seesaws reconstructed from the observed data apparently reproduce the AO.