Tut 3 answers.
==============

1) My routine is in ims_linfit.

2) a)
		                      T
		sigma_T = -------------------------
		           sqrt(delta_t * delta_nu)
T_sys = 30 K
delta_t = 1/2 hour, = 1800 s
delta_nu = 32e6 Hz
Therefore
		                     30
		sigma_T_off = -------------------
		               sqrt(1800 * 32e6)

		            = 125 microK.


   b) This is just propagation of errors.

	T_nett = T_on - T_off,

thus
	sigma_T_nett = sqrt(sigma_T_on**2 + sigma_T_off**2).

But if T_src << T_sys, T_on is approx = T_off, and (provided delta_t is the same for on and off), sigma_T_on is approx = sigma_T_off. Thus

	sigma_T_nett = sqrt(2) * sigma_T_off
	             = 177 microK.


   c)
	        2k
	S_Jy = ----- * T_nett
	        A_e
hence
	              2k
	sigma_S_Jy = ----- * sigma_T_nett
	              A_e

	           = 174 microJy.

3) a) i) Because only a power law gives a straight line on a log-log plot.
      ii) Fitted spectral index (the slope of the line) = 0.75.
      iii) This value is typical for a synchrotron source.

   b) The polarization fraction = I / sqrt(Q*Q + U*U + V*V)
                                = 0.020 at 6 cm.

   c) No, because no value is significantly larger than the uncertainty of +-0.05.

   d)
	Lambda (cm)	 angle (deg)
	 6		 49.5  +- 2.5
	10		 41.4  +- 1.4
	21		 -2.37 +- 0.54
	30		-62.01 +- 0.27

   f)
	intercept & slope: 54.7, -0.130