Each tarball contains the Markov Chain files for a given dataset/model combination.
The chains were run using an eight-process job; each process wrote its own output
file.  This format was retained for later efficiency.

Each file contains the following columns:

col     param
---	----------------------------------------------------------------------------
1       weight
2       -ln(likelihood)
3       \Omega_b h^2
4       \Omega_c h^2
5       \theta (cosmomc format)
6       \tau (except in lcdm_xestep where it is x_e)
7       \Omega_K
8       f_\nu or N_{rel} for models which do not contain both
9       w
10      n_s (defined at k=0.05 in lcdm_run and k=0.002 in lcdm_run_tens)
11      n_t
12      dn_s/d\ln k
13      ln (10^10 A_s) (defined at k=0.05 except models with tensors where k=0.002)
14      r (defined at k=0.002 always)
15      A_{SZ}
16      \Omega_L
17      age/Gyr
18      \Omega_m
19      \sigma_8
20      z_{rei}
21      ZERO (cosmomc usually stores r at l=10 here for relevant models)
22      H_0
23      \beta_{SDSS}
24      b_{SDSS}
25      z_{bg}


