**4.3. Harmonics of CBR Anisotropies**

so I will only briefly mention it for completeness. The positions of the acoustic peaks are particularly sensitive to and , and even low accuracy data available at present lead to a meaningful constraint on a combination of and .

The first acoustic peak appears at *l* =
(the distance
to the last-scattering surface)/(the sound horizon)
(Hu & Sugiyama 1995).
Its position *l*_{1} is approximated as

for the parameter range that concerns us.
This means that the position of the acoustic peak is about
*l* 220 if
+
= 1, but it
shifts to a high *l* as
^{-1/2} if
= 0.
On the other hand, there is little power to determine
separately from , unless full
information of *C*_{l}
is used.
The harmonics *C*_{l} measured at small angles revealed the
acoustic peak
(Scott et al. 1996),
and its position favours a universe not far from flat
(Hancock et
al. 1998).
More exhaustive analyses of
Lineweaver
(1998) and
Efstathiou et
al. (1999)
show a limit
+
/2 > 0.52
(1). (The contours of the
confidence level
fall approximately on the curve given by (15) with *l*_{1}
= constant.)
This means that a zero
universe is already marginal,
when combined with
from other arguments.
If a flat universe is chosen from CBR, a non-zero
will be compelling.