Dear Editor,

we first would like to thank the referee for the fast reading, and the
constructive remarks on the manuscript. To address the suggestions and
questions raised in the review, we would like to go through them in their
original order:

1a. (Specification on the lens size)
At the end of the first sentence in section 2 (Experimental Setup), we added 
", resulting in an effective lens aperture of 1.2 mm." 
to indicate the effective lens diameter. The lens diameter itself is larger,
(around 6mm), but we only use the central part.

1b. (Clarification on integration times)
The quoted 4 minute intervals for measurement coincides with the integration
time; this was not so clear in the main text. Therefore, just before the old
eq. 6 (new equation 7), we replace the text 
"... is shown in 4 minute time intervals." by
"... is shown for consecutive integration intervals of 4 minutes."

2. Indeed, the transverse coherence length of the sun is smaller than the
effective size of the coupling lens. However, the single transverse mode fiber
ensures that we have full spatial coherence after the fiber. The limited
spatial coherence of sunlight before the fiber should result in a reduced
coupling efficiency, but not affect the temporal structure of the collected
light. We do not expect a reduction of the temporal g(2) seen behind the
single mode fiber due to this limited coupling efficiency.


3. We appreciate the suggestion to discuss the SNR a bit more in detail. To do
so quantitatively, we added a paragraph (around new equation 6) right
after the presentation of the solar g(2):

"The observed ratio of the photon bunching excess (g^(2)-1) to the noise
given by the standard deviation Delta evaluated from propagated Poissonian
statistics in our experiment is SNR=(g^(2)-1)/Delta=13.8. This is
slightly smaller than the signal-to-noise ratio predicted in (hbt:1974),

  SNR=1/2r_s\sqrt{\tau_j T}~21,

where r_s~300000/s is the counting rate on a single detector, \tau_j the time
bin width for the pair histogramming (40\,ps), and T=4 minutes the integration
time for one measurement. The discrepancy could be explained by a varying r_s
due to  fast weather condition changes over T."



4.Preliminary estimations of the single line emission of gamma Velorum seem to
suggest that we could expect single photon detection rates on the order of 
10^ 6/sec from a 8" telescope. If we were to compare this to the signal
rates for sunlight presented in this paper, we should be able to observe a
signature in a very short time. However, we hesitate to make such a statement
in the manuscript, because we do not know what coupling efficiency from
starlight in to the filter system will be achievable.


5 . We certainly agree that a better knowledge of the time response of the
detectors would help for making quantitative statements on decoherence effects
even with the detectors we have now. However, this turned out to be
complicated, because we currently can not test the detector response with a
sufficient time resolution at the green wavelength. The detector measurements
we present in figure 3 had to be made at 810nm, where the behavior of the
detectors may be significantly different due to the rapidly changing
absorption length in silicon. We are trying to work on this, but have no
results at the moment that would help for this manuscript.                


6. Typo - we corrected it in the second last paragraph.


With this, we hope to have addressed the suggestions by the referee, and look
forward for your reply.

Best Regards on behalf of all authors,

Christian