Dear Editor,

first of all, we would like to thank the referees for their careful reading,
their positive comments, and the constructive suggestions on how to improve the
manuscript. Here our replies to the issues raised:

Comments of Referee A

The main comment of Referee A is that the paper is "narrow-fielded" and
therefore more justifications are needed for its acceptance to PRL. We want to
note that the problem of temporal shaping of photons is also essential in
cavity QED experiments, in which there is a need to transfer photons between
several atoms in cavities, simulating quantum networking (Nature 484, 195
(2012) and others). While spatial mode matching between distant cavities is
very efficient, the frequency/temporal profiles of photons must be controlled
to boost the communication efficiency.  It had also been shown that even in
the absence of atoms, an exponentially rising coherent pulses lead to higher
energy storage in optical resonators (G. Leuchs, unpublished). Moreover if
spatial mode overlap is further increased (e.g. by using deep parabolic
mirrors) the atom+mirror system would look like an  atom in half-cavity thus
resembling standard QED systems. To address the range of interest, we like to
point out that our experiment has a good overlap with theoretical predictions,
which are based on the full quantum mechanical treatment of atom-field
interactions.  Therefore, any research group which deal with similar problems
in ion traps and single molecules and quantum dots can adopt the presented
theory which makes quantitative predictions.

Also, the temporal pulse shaping may be relevant for researchers on quantum
storage in atomic ensembles, since our work presents the simplest of all
cases, namely a single two-level atom in the "strong" field of a weak coherent
field mode, and may serve as a reference for comparison there.

We have adopted the suggestions of comments of referee A in this direction and
added explicit references to cavity QED experiments in the introductory part,
and in the summary.

We also added an explanation of the open/filled data symbols in the caption of
figure 5.

As for the length concerns, we seemed to be within the page limit, and we
wanted to keep the main physics setup figure since we believe it
gives a quick overview over the physical configuration of the experiment.
We did streamline the details for the pulse generation a little, singe many
details can be found in reference 17.


Comments of Referee B 

A group of comments of Referee B takes issue with our statement about
observation of Rabi oscillations in the abstract and the supporting data in
Figure 5. We agree that the Rabi oscillations are probably not as clear as we
stated e.g. in the abstract, certainly for the presented data with
exponential excitation profile. But we still hold to the claim that the square
pulse leads to quite clear oscillatory population, with  a visibility around
50%. Rabi oscillations with same visibility can be seen for exponential pulse
as well, but for smaller temporal width (5 ns).

The comment of the referee of an undersampled trace for the square pulse in
figure 4 is understandable, but a result for us only showing a subset (only
every 3rd point) of the data we took in order to make the graph readable. If
we display all points, the graph gets in our view too cluttered, but stays at a
visibility of about 50%. We are convinced that a part of the dynamics of the
atom in the pulsed light can be well understood as Rabi oscillations,
particularly for 
pulse durations shorter than the natural life time. However, the purpose of
this work was not to present Rabi oscillations, so we reworked the manuscript
in the following ways to focus more on the excitation part:

1. We removed the statement about Rabi oscillations from the abstract. The
focus is the excitation of the atom with a low photon number, and for this
parameter regime we would not see the traditional oscillatory behavior for an
exponentially rising pulse in the first place.

2. We quote the visibility of the Rabi oscillations in figure 4 explicitly.

3. We added a graph as supplementary material that shows an oscillatory
aspect of the atomic population for a shorter pulse  (5 ns) more clearly for
the exponentially rising envelope.

4. We removed the comments referring to the connection between Rabi
oscillations and possibilities for switching, since the relevance of these
schemes for the work presented in this manuscript is rather indirect.


The next comment is about the deviation of the experimental points from the
model prediction for t>0 for square pulse curve in Fig. 4. The reason to that is
that the edges of the square pulse are not perfect, since the filter etalons
we had available cut out the required optical sideband had a ringdown time of
approximately 2 ns. This imperfection makes a square pulse rather
"trapezoidal", and changes the phase at the end of the Rabi oscillation when
compared to the square envelope of the theoretical model. For the parameters
in figure 4, the population oscillation was already on the down hill slope, so
small changes in the phase result in significant change in the
population. Unfortunately, we don't have a measure that would capture the
total Rabi phase expected for our pulse with a high enough accuracy that we
could predict the Rabi phase at the end of the pulse accurately enough to be
really meaningful.


Minor comments of Referee B.

1. The comment about the square pulse edges is absolutely valid and we
explained the source of imperfections above. This will of course alter the
results for small pulse durations and it is clearly seen in top left curve of
Fig. 5 for the square pulse. However we feel that the main result (Fig. 6) is
unaffected by that imperfection, since the  temporal width of the pulse is
much larger than the characteristic rise/fall time of the pulse. Nevertheless,
we agree to the Referee's comment about this issue and added brief discussion
in the experimental section. 

2. Motivation of deviations of experimental points from theory in Fig. 6 
due to thermal motion of atoms in a trap. We added this comment explicitly in
the text; however in Fig. 4, the theory fits our data quite well for different
pulse parameters. At the moment, we have to conclude that either  we had
different temperatures of the atom for the different experimental runs, or
there are other effects responsible for it we have not fully understood yet.

3. Choice of the average photon numbers for data in Fig 6.
The choice for the photon number and pulse length combination was motivated by
the availability of measurement data at low photon numbers with the "fair"
combination of time constants for both pulse shapes explained in the
paragraph before the summary. We chose a low photon number for figure 6 to
avoid complications with the nonlinear response, and also to indicate the
degree of excitation we can accomplish with the lowest photon number pulse we
could measure with a reasonable signal to noise ratio.
We did not find a statistically significant increase of the relative advantage
of a exponential pulse with the photon number, as perhaps suggested by the
Referee, nor would the simulation support this - to the contrary. While we
don't have well-matched experimental data to show this directly, the ratio of
Pe(exp) and Pe(sqare) in the numerical data e.g. for the 15ns case decreases
monotonously from about 2.1 to about 0.7 with <N>. However, we understand that
one can easily come to the visual interpretation highlighted by the Referee from
the appearance of figure 5, possibly due to the logarithmic scale we had to
use for <N> to capture its wide relevant range. Therefore, we added a clause
in the caption of figure 5 to mention the change in ratios explicitly.

We hope to have addressed the issues raised by the referees and look forward
for your reply.

At the moment, we have included the supplementary material for easy
assessment, but we can easily take it as a small separate pdf file. We also
are open to any format that you would recommend to us for this.

Best Regards on behalf of all authors,

Christian Kurtsiefer