Results

The results of the fit are summarised in Table.2. The combination of the two runs is done by the simultaneous fit. The first line corresponds to pure V-A SM fit. In the second line the tensor and in the third the scalar terms are added into the fit. All the errors presented are from the "MINOS" procedure of the "MINUIT" program [14] and are larger than the gaussian ones. At present, we estimate an additional systematics error in $\lambda_{+}$ to be $\pm 0.002$. The estimate is done by comparing two runs, which differ a lot in amount of matter in the beamline and detector configuration and by varying cuts, cell size during the fit of the Dalitz plots etc.

Table 2: Results of the fit.

	1999	2001	1999+2001
$\lambda_+$	$~~0.0284^{+0.0036}_{-0.0036}$	$~~0.0313^{+0.0020}_{-0.0020}$	$~~0.0306^{+0.0018}_{-0.0018}$
$\lambda_+$	$~~0.0284^{+0.0036}_{-0.0036}$	$~~0.0313^{+0.0020}_{-0.0020}$	$~~0.0306^{+0.0018}_{-0.0018}$
$f_T = F_T/f_+(0)$	$-0.0443^{+0.1091}_{-0.1036}$	$-0.0489^{+0.0856}_{-0.0797}$	$-0.0444^{+0.0678}_{-0.0648}$
$\lambda_+$	$~~0.0267^{+0.0047}_{-0.0052}$	$~~0.0310^{+0.0022}_{-0.0025}$	$~~0.0299^{+0.0019}_{-0.0020}$
$f_S = F_S/f_+(0)$	$-0.0355^{+0.0648}_{-0.0283}$	$-0.0180^{+0.0341}_{-0.0249}$	$-0.0152^{+0.0249}_{-0.0197}$
$\chi^2/\mbox{ndf}$	1.40	1.30	1.37
$N_{\mbox{bins}}$	168	220

The comparison of our results with the most recent $K^{\pm}$ data [1,3,4] shows that our statistics, at present, is the highest in the world and the statistical errors somewhat smaller than in [1,3] and comparable with [4]. We do not confirm the observation of a significant $f_{S}$ and $f_{T}$ in [1]. Our data is in a good agreement with [3,4] and with the theoretical calculations for $\lambda_{+}$: $\lambda_{+}=0.031$ [15], done in the context of the chiral perturbation theory.