Why thin wire array behaves like a plasma medium? (Metamaterial of negative permittivity)

Anyone who had ever read Sir John Pendry's paper "Extremely Low Frequency Plasmons in Metallic Mesostructures" (Phs. Rev. Lett. 76,  4773 (1996)) may feel the arguement Pendry used for deriving the plasmon behavior is a little bit unclear. For example, Pendry replaced the canonical momentum (containing the qA term) of the electron inside the wire by an "effective kinetic momentum", which means the "effective electron" has a very large mass. The ambiguity can appear when integrating the H field to obtain a vector potential. Besides, the gauge freedom of the vector potential implies an additional uncertainty of the effective momentum. 

To derive the effective plasma frequency correctly, one should notice the following two facts. First, the wire cross secton occupies a small portion of a unit cell, thus the effective concentration of charge in the medium is much lower than that in the "pure metal" case. Second, the wire array has strong magnetic effects (large inductance, large magnetic field around the wire), which is lacking in the original "pure metal" medium.

I believe I have found a more transparent argument, which I want to share here. The following two figures include the details.

If you cannot see the figures, please connect to the following website (chinese interface):

http://pgluan2010.wordpress.com/2010/04/04/%e7%82%ba%e4%bd%95%e5%b9%b3%e8%a1%8c%e9%87%91%e5%b1%ac%e7%b7%9a%e9%99%a3%e5%88%97-wire-array-%e5%8f%af%e4%bb%a5%e8%a2%ab%e8%a6%96%e7%82%ba%e7%ad%89%e6%95%88%e9%9b%bb%e6%bc%bf-%e7%ad%89%e9%9b%a2/