it is a plot of the power spectral density v/s frequency. It is a plot obtained from finite time samples of data. After finding the auto correlation of the time sequence, we take the fourier transorm (DFT) of the sequence to obtain the power spectral density. I do not have a picture right now, but i will attach it as soon as i can get access to a scanner.
U have to seek local maximums in your periodogramm. The each local maximum corresponds to main frequency in periodic component in your signal. But if u analysing randon signal with additive periodic components u have to average several signal bloks for obtaining one periodogramm (for eliminating of finite-length random signal realisation on your spectralestimation).