We consider Gauss quadrature formulae
Q
n
{Q_n}
,
n
∈
N
n \in {\mathbf {N}}
, approximating the integral
I
(
f
)
:=
∫
−
1
1
w
(
x
)
f
(
x
)
d
x
I(f): = \smallint _{ - 1}^1w(x)f(x)\;dx
,
w
=
W
/
p
i
w = W/{p_i}
,
i
=
1
,
2
i = 1,2
, with
W
(
x
)
=
(
1
−
x
)
α
(
1
+
x
)
β
W(x) = {(1 - x)^\alpha }{(1 + x)^\beta }
,
α
,
β
=
±
1
/
2
\alpha ,\beta = \pm 1/2
and
p
1
(
x
)
=
1
+
a
2
+
2
a
x
{p_1}(x) = 1 + {a^2} + 2ax
,
p
2
(
x
)
=
(
2
b
+
1
)
x
2
+
b
2
{p_2}(x) = (2b + 1){x^2} + {b^2}
,
b
>
0
b > 0
. In certain spaces of analytic functions the error functional
R
n
:=
I
−
Q
n
{R_n}: = I - {Q_n}
is continuous. In [1] and [2] estimates for
‖
R
n
‖
\left \| {{R_n}} \right \|
are given for a wide class of weight functions. Here, for a restricted class of weight functions, we calculate the norm of
R
n
{R_n}
explicitly.