Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/irlab/doc/progress.ps.gz Äàòà èçìåíåíèÿ: Thu Jul 28 18:22:39 2005 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 02:39:28 2007 Êîäèðîâêà:

An ARNICA Progress Report
L. K. Hunt & F. Mannucci
C. A. I. S. M. I. ­ C. N. R.
Largo E. Fermi 5, I­50125 Firenze, Italy
Electronic mail: hunt@arcetri.astro.it, filippo@arcetri.astro.it
ABSTRACT
This report describes technical results obtained with the Arcetri
near­infrared camera, ARNICA, mounted at three different telescopes
from 1995 to 1998: TIRGO, the Nordic Optical Telescope, and the Vat­
ican Advanced Technology Telescope. For each telescope, we present
estimates of the camera+atmosphere efficiency, point­ and extended­
source sensitivity, telescope emissivity, and mean atmospheric extinc­
tion coefficients. We also use data from January, 1995 to January, 1998
to analyze ARNICA's bad pixels and their change with time.

1. Introduction
More than five years have passed since
the Arcetri near­infrared (NIR) camera, AR­
NICA, was first mounted at TIRGO in 1992
(Lisi et al. 1993, Lisi et al. 1996). During
the time that the camera has been in oper­
ation, it has acquired more than 60 Giga­
bytes of data (about half a million frames)
and has been mounted at five different tele­
scopes.
Here we present a comparative analysis
of data from ARNICA mounted at three
different telescopes from 1995 to 1997: 1) the
1.5­m f/20 Gornergrat Infrared Telescope 1
(TIRGO); 2) the 2.56­m f/11 Nordic Opti­
cal Telescope 2 (NOT); 3) the 1.8­m f/9.5
Vatican Advanced Technology Telescope 3
(VATT). Technical results from the initial
campaigns (1992­1994) at TIRGO and the
William Herschel Telescope (WHT) have
already been reported in Hunt et al. 1996,
and a series of technical reports (Hunt et
al. 1994a, 1994b, 1994c). The camera, AR­
NICA, relies on a NICMOS3 256\Theta256 1--
2.5 ¯m HgCdTe scientific grade array, and
provides a 4\Theta4 arcmin FOV with 1 arc­
sec pixels at TIRGO, and 2\Theta2 arcmin with
0.5 arcsec pixels at VATT and NOT.
In the following we report estimates for
system efficiency, sensitivity, emissivity, and
mean atmospheric extinction coefficients at
the different telescopes. We also analyze
ARNICA's bad pixels, and discuss the grad­
ual temporal degradation of the cosmetic
quality of the array. The latest bad pixel
mask is derived from ARNICA observations
obtained in January, 1998.
2. Methods and results
We have analyzed ARNICA data for 23
photometric nights, most of which are in­
cluded in the database for the ARNICA
1 TIRGO (Gornergrat, Switzerland) is operated by
CAISMI­CNR, Arcetri, Firenze.
2 NOT is operated on the island of La Palma jointly
by Denmark, Finland, Norway, Sweden, in the
Spanish Observatorio del Roque de los Muchachos
of the Instituto de Astrofisica de Canarias.
3 VATT (Mount Graham, Arizona) is composed of
the Alice P. Lennon Telescope and the Thomas J.
Bannan Astrophysics Facility, both operated by the
University of Arizona (Steward Observatory) and
the Vatican Observatory Research Group.
Standard Star Network (Hunt et al. 1998).
Table 1 lists the diameters and pixel scales
of the three telescopes considered here. Sen­
sitivity calculations, flux conversions, and
filter characteristics are as given in Hunt
et al. 1994b) 4 . The only exception is that
here point­source limiting magnitudes for a
given night are derived using an effective
aperture defined by 4 times the Gaussian oe
(1.7 \Theta FWHM) of the median point­spread
function (PSF) measured on that night. In
all sensitivity estimates, the collecting area
is determined from the primary mirror di­
ameter; neither the loss in collecting area
from the central hole in the primary nor
the secondary mirror obscuration was taken
into account. This means that, especially
for NOT and VATT, the values in Table 2
are underestimated.
Table 2 reports the results our anal­
ysis for the different telescopes, both in
terms of nightly variations, and of global
site/telescope means, averaged over the var­
ious observing campaigns. The various
columns report complete nightly medians
that typically involve hundreds of frames.
The only exception is the VATT campaign
for which we have only mean values of the
PSF randomly sampled over the run.
2.1. Efficiency, sensitivity, and see­
ing
The efficiency j is a global value, and
takes into account losses from the atmo­
sphere, the telescope, and the camera. The
highest efficiencies were measured at VATT,
and are probably due to the lower number
(two) of reflecting surfaces in the optical
train. The efficiencies at NOT and those
at TIRGO in the 1996 run are comparable
and higher than those measured in the early
runs (1992­1994) at TIRGO. The efficien­
cies in the April, 1997 TIRGO campaign are
degraded relative to the year before, and
are very slightly lower than the 1994 val­
ues; this degradation may be attributable
to the need for aluminization which was
performed in Summer, 1997.
4 F 0 = 1560, 1030, 645 Jy, in J , H, and K, respec­
tively.
– = 1.25, 1.65, 2.20 ¯m in J , H, and K, respec­
tively.
\Delta– = 0.3, 0.3, 0.4 mum J , H, and K, respectively.
2

Table 1
Characteristics of Telescope+ARNICA
Telescope Diameter (m) f/ number Pixel (arcsec)
TIRGO 1.50 20 0.966
NOT 2.56 11 0.546
VATT 1.83 9.5 0.45
The point­source sensitivities should scale
with telescope diameter in a background­
limited regime, if background and seeing
are similar. This is clearly not the case for
our data set. TIRGO's rather high J­ and
H­band backgrounds make it relatively in­
efficient for 1 ¯m work, while its low back­
ground in K makes it comparable to tele­
scopes with as much as three times its col­
lecting area, when the seeing is good as in
the 1996 run. NOT and VATT sensitivities
scale rather well, except for the K band,
where the slightly lower VATT background
makes it comparable to NOT even with half
its collecting area.
The median seeing FWHM of the 1996
TIRGO campaign is slightly better than
2 arcsec, that is to say roughly 2 pix­
els; therefore, this value is probably an
upper limit, as narrower PSFs than these
are difficult to measure with ARNICA's 1­
arcsec pixels at TIRGO. The 1997 run at
TIRGO suffered from worse seeing (3 arcsec
FWHM), and the point­source sensitivities
for this period are consequently lower.
The seeing at NOT is typically good at
1.2--1.4 arcsec in K, although occasionally
the seeing can be very good; 20­30% of
each night was characterized by sub­arcsec
FWHMs. VATT data are not as detailed,
but the mean seeing at Mt. Graham ap­
pears comparable to that of La Palma.
2.2. Telescope emissivity and K­band
background
The measurement of telescope+instrument
emissivity is given by the ratio of two quan­
tities: 1) the non­atmospheric contribu­
tion to the measured background (NAB),
and 2) the flux density from a blackbody.
Typically, for the blackbody measurement
(BB), one observes an unlit dome or the
closed mirror covers. The NAB can be de­
rived from a regression of sky versus air­
mass, if we assume that the true sky back­
ground is proportional to atmospheric path
length; any atmospheric contribution from
a component that does not vary with air­
mass would be attributed to NAB. In the­
ory, the intercept of such a regression is the
background at zero airmass, that is to say
the emission of the telescope which we as­
sume to be the sole contributor to NAB.
This value should also be comparable to the
difference of the minimum (presumably at
unit airmass) observed sky background and
slope of the above regression. If we have
measurements at airmasses (am) 1 and 2,
then the NAB can be derived with the sim­
pler expression, NAB=[am1 ­ (am2­am1)].
In all cases, emissivity is therefore NAB/BB.
Figure 1 shows the regressions obtained for
the three different telescopes, together with
the best­fit regression lines.
TIRGO is characterized by a low emis­
sivity, 12%, as expected because of its
infrared­optimized design, and is similar to
that measured at UKIRT (12.5% for the
three telescope reflecting surfaces). Both
optical telescopes, on the other hand, have
rather high emissivities: 35% (NOT) and
40% (VATT). In all cases, the two methods
-- or consistency checks -- described above
(intercept; minimum sky minus the slope)
agree to within 1%. The method of simply
measuring the two airmasses performed at
VATT gives an emissivity of 40%, which is
similar to results from the other two meth­
ods: 43% and 41%, respectively. We note fi­
nally that (for all telescopes) these are lower
limits to the emissivity because the black­
body measurement was always performed
3

Table 2
ARNICA Efficiencies, Sensitivities, and PSF's
j Extended Source Point Source Sky FWHM
Site (mag arcsec \Gamma2 ) a (mag) b (mag arcsec \Gamma2 ) (arcsec)
Date J H K J H K J H K J H K J H K
NOT
95aug08 0.18 \Delta \Delta \Delta 0.36 20.5 \Delta \Delta \Delta 18.0 20.2 \Delta \Delta \Delta 18.0 16.3 \Delta \Delta \Delta 11.8 1.51 \Delta \Delta \Delta 1.21
95aug09 0.18 \Delta \Delta \Delta 0.36 20.4 \Delta \Delta \Delta 17.9 20.2 \Delta \Delta \Delta 18.0 16.2 \Delta \Delta \Delta 11.8 1.44 \Delta \Delta \Delta 1.11
95aug10 \Delta \Delta \Delta 0.38 0.36 \Delta \Delta \Delta 19.2 17.9 \Delta \Delta \Delta 18.6 17.4 \Delta \Delta \Delta 13.7 11.7 \Delta \Delta \Delta 1.99 1.93
95aug11 0.16 0.37 0.34 20.0 19.2 17.6 20.1 19.3 17.5 15.5 13.7 11.1 1.12 1.09 1.26
MEAN 0.17 0.37 0.35 20.3 19.2 17.8 20.2 19.0 17.7 16.0 13.7 11.6 1.36 1.54 1.38
TIRGO
96jan16 0.17 0.31 0.33 19.8 18.7 18.3 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta 15.0 12.9 12.6 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta
96jan17 0.16 0.31 0.34 19.6 18.5 18.3 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta 14.5 12.4 12.4 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta
96jan18 0.18 0.32 0.34 19.7 18.5 18.3 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta 14.7 12.5 12.5 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta
96jan19 0.17 0.33 0.34 19.7 18.6 18.4 18.6 17.4 17.4 14.8 12.6 12.6 1.89 2.05 1.74
96jan20 0.18 0.33 0.34 19.7 18.6 18.4 18.6 17.6 17.5 14.7 12.7 12.7 1.87 1.73 1.61
96jan21 0.18 0.33 0.34 19.3 18.7 18.4 18.1 17.6 17.3 13.9 12.8 12.6 2.12 1.87 1.86
MEAN 0.17 0.32 0.34 19.6 18.6 18.3 18.4 17.6 17.4 14.6 12.7 12.6 1.96 1.88 1.74
NOT
96aug31 0.18 0.34 0.32 20.1 19.2 17.8 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta 15.6 13.9 11.7 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta
96sep01 0.19 0.34 0.33 20.2 19.2 17.8 20.1 19.0 17.7 15.8 13.8 11.7 1.43 1.40 1.39
96sep02 0.18 0.33 0.33 20.2 19.1 17.8 20.2 19.0 17.7 15.8 13.8 11.7 1.20 1.43 1.45
96sep03 0.18 0.34 0.33 20.4 19.3 17.9 20.0 19.1 17.6 16.1 14.0 11.7 1.65 1.49 1.57
96sep04 0.17 0.33 0.32 20.3 19.3 17.9 20.1 19.4 17.6 16.1 14.1 11.9 1.50 1.08 1.59
MEAN 0.18 0.34 0.33 20.3 19.2 17.9 20.1 19.1 17.6 15.9 13.9 11.7 1.44 1.35 1.50
VATT
96dec15 0.25 0.42 0.40 19.7 18.5 17.4 19.6 18.5 17.4 15.5 13.4 11.8 1.60 1.44 1.46
96dec16 0.24 0.40 0.39 19.7 18.5 17.5 19.6 18.5 17.5 15.6 13.4 12.0 1.60 1.44 1.46
96dec17 0.23 0.40 0.39 19.8 18.4 17.3 19.7 18.4 17.3 15.8 13.3 11.6 1.60 1.44 1.46
96dec18 0.25 0.40 0.40 19.8 18.2 17.5 19.7 18.2 17.5 15.7 12.8 12.0 1.60 1.44 1.46
96dec20 0.24 0.42 0.40 19.8 18.6 17.6 19.7 18.7 17.6 15.8 13.7 12.1 1.60 1.44 1.46
MEAN 0.24 0.41 0.40 19.7 18.5 17.5 19.6 18.5 17.5 15.7 13.3 11.9 1.60 1.44 1.46
TIRGO
97apr13 0.13 0.26 0.30 19.7 18.7 18.3 18.0 17.1 16.7 15.0 13.0 12.6 3.21 2.89 3.02
97apr14 0.13 0.27 0.29 19.9 18.9 18.4 18.7 17.7 17.0 15.6 13.5 12.8 2.10 2.19 2.36
97apr15 0.13 0.26 0.29 19.8 18.9 18.3 17.9 17.1 16.8 15.3 13.4 12.6 3.86 3.34 2.80
MEAN 0.13 0.26 0.29 19.8 18.8 18.3 18.2 17.3 16.8 15.3 13.3 12.7 3.06 2.81 2.73
a Calculated at 3 oe, in 60 sec.
b As a, and assuming a virtual aperture of 1.7 FWHM (Cols. 14--16).
4

Fig. 1.--- Observed K­band background
surface brightness (image median) versus
airmass. The background surface bright­
ness is given in ph/s/m 2 /arcsec 2 . The data
are from the night when emissivity was
measured, and the dotted lines show the
best­fit regression described in the text.
at the beginning of the night, when the
ambient temperature is presumably warmer
than it is later in the night.
Our estimate of the non­atmospheric back­
ground at NOT (1.77 \Theta 10 4 ph/s/m 2 /arcsec 2 )
is similar to that at VATT (1.84 \Theta 10 4 )
but slightly higher than that at TIRGO
(1.35 \Theta 10 4 ). The higher levels are almost
certainly due to similarly high telescope
emissivity, although the temperatures on
the two runs were somewhat different (+8C
at NOT, \Gamma7C at VATT). The pure sky
background emission (at unit airmass) ranges
from 0.2 \Theta 10 4 to 0.7 \Theta 10 4 , with the lowest
value at VATT and the highest at NOT.
Emission from OH molecules is thought to
be the dominant component of NIR airglow
up to 2.3 ¯m (Oliva & Origlia 1992), and
indeed non­thermal models of NIR airglow
approximate reasonably well the observed
spectra (McCaughrean 1988). Such emis­
sion originates at 90 km above the Earth's
surface, so the different values should not
reflect altitude dependence; they more likely
reflect the night­to­night or site­to­site vari­
ations in the local OH properties.
Finally, we note that the atmospheric
contribution to the observed background
is estimated to be 28% at NOT, 20% at
TIRGO, and 9% at VATT. Clearly, the
observed K­band background at VATT is
dominated by the telescope emission. For a
zero­emissivity telescope at the Gornergrat
or Mt. Graham, we derive a background
emission of around 14.4--14.5 mag/arcsec 2 ,
with a brighter 13.1 mag/arcsec 2 at La Palma
for our nights.
2.3. Atmospheric extinction coeffi­
cients
Atmostpheric extinction in each band
was measured by combining all the nights
in a given observing run, and assuming that
the extinction is constant from night to
night, but that the zero point at unit air­
mass varies. For an observation of the j th
calibrating star on the i th night, we have:
ZP obs
(i;j) = ZP am=1
(i;j) + k (am \Gamma 1). For each
run in each band, we performed a simulta­
neous fit over the n nights of k and the n
values of ZP am=1
i . A few of these fits are
shown in Fig. 2, and Table 3 gives the ex­
tinction coefficients for the various runs at
5

Fig. 2.--- K­band ZP (i;j) \Gamma ZP i vs. airmass.
Data point are given by the night number
(i), and the dotted lines show the best­fit
regressions described in the text. The val­
ues of the slopes are given in Table 3.
the different telescopes. Results from early
ARNICA commissioning runs (1992­1993)
and mean photometer values (Hunt et al.
1987) are also included in the table.
Table 3 shows that, although the K­band
sky emission is substantially higher at La
Palma than at Mt. Graham or the Gorner­
grat, its extinction properties are similar:
the attenuation of zero point with airmass
is, on average, ­0.04 mag/airmass in K at
all sites.
3. Bad pixels
An operational definition of a bad pixel
(bp) is a subject of controversy. Some con­
sider bad pixels those which ``stand out'' in
a direct image, that is pixels whose count
levels differ significantly from their neigh­
bors; others consider bad pixels those which
stand out after the flat­field process, or
those which do not behave according to the­
oretical expectations for the noise. Indeed,
a complete and rigorous analysis of the bad
pixel behavior in hybrid NIR arrays would
be the subject of at least a paper, and per­
haps a series of them. Nevertheless, for con­
venience, we adopt here the former defini­
tion of a bp: bp's are singled out by their
anomalous behavior relative to the pixels in
the neighborhood in the raw images. The
definition of ``neighborhood'' should be such
that clearly anomalous clusters of bad pix­
els are distinguished by the ``good'' pixels
surrounding them.
We used a procedure similar to that de­
scribed in Hunt et al. (1994c), in which the
original high signal­to­noise image is me­
dian smoothed, subtracted from the origi­
nal, and the difference image histogrammed.
The histogram is then clipped until it ap­
proximates a symmetric distribution with
no outliers; the clipped pixels are defined
as ``bad'' and inserted into the bp mask
for that date. One difference in the cur­
rent procedure with respect to 1993 is that
now the initial smoothing window had to
be larger (10 as opposed to 5 pixels) be­
cause of the growth of bp pixel clusters (in
particular between 1995 and 1996). Also,
the process now needs to be iterated as, be­
cause of the larger smoothing window, one
pass frequently fails to detect all the bp's
that degrade the histogram of the difference
6

Table 3
Extinction Coefficients
Site Date No. Nights k J a kH a kK a
TIRGO 1992 December 8 ­0.066 (0.034) ­0.061 (0.035) ­0.053 (0.039)
1993 February 4 ­0.131 (0.057) ­0.048 (0.040) ­0.021 (0.050)
1996 January 6 ­0.042 (0.023) ­0.006 (0.015) ­0.060 (0.029)
1997 April 3 ­0.092 (0.033) 0.000 (0.022) ­0.117 (0.040)
MEAN 21 ­0.083 (0.038) ­0.029 (0.030) ­0.063 (0.040)
TIRGO 1986­1987 b 5 ­0.126 (0.040) ­0.060 (0.019) ­0.076 (0.026)
NOT 1996 September 5 ­0.057 (0.022) 0.000 (0.014) ­0.042 (0.019)
VATT 1996 December 5 ­0.067 (0.020) ­0.007 (0.012) ­0.040 (0.014)
a Units are mag airmass \Gamma1 . Errors are given in parentheses.
b Mean photometer values taken from Hunt et al. 1987. Standard deviations are given in
parentheses.
image. When we had more than one image
available, masks were determined from each
image and compared, and final masks were
those which successfully eliminated all bp's
in all of the images from that date.
We have analyzed mask data from Jan­
uary, 1995, to January, 1998, for a total of
five data sets. The 97sep24 mask was ac­
quired by G. Comoretto and R. Maiolino,
and the 98jan24 mask by P. Ranfagni, and
both were kindly made available to us. The
results of the bp selection procedure are
given numerically in Table 4. These can be
compared with values from June, 1993 (in
which the regions are not made contiguous
for interpolation), also given in the table.
Generally speaking, the cosmetic qual­
ity of ARNICA has degraded slightly from
0.9% in 1993 to 1.2% (Col. 3) five years
later; nevertheless ARNICA's array is still
in reasonably good shape. We note how­
ever several potentially puzzling points that
have emerged from our analysis, and then
examine each in detail:
1. The bad pixel list does not grow mono­
tonically with time, and not all pixels
in an earlier list are contained in the
subsequent one.
2. The number of bad pixels defined from
a given image depends (in four of five
cases) on the count level; the higher
the level, the fewer the bad pixels.
3. Two bad columns (128), (256) are
clearly present in only two of the five
mask images, and their presence does
not depend on count level or other pa­
rameters that we know of.
4. The latest two images (97sep, 98jan)
exhibit double­peaked histograms
(when the smoothed image is sub­
tracted as described above) that seem
to depend on specific quadrants and
whose characteristics change with count
level.
One would logically assume that previ­
ous bad pixel lists should be subsets of later
ones, and that the true detector bad pix­
els should increase with time. Such as­
sumptions are not warranted in the data
we have analyzed here. First, inspection
of Table 4 (either Col. 2 or 3) shows that
the number of bp's decreases between 95jan
and 95nov. This is almost certainly due
to the increase of the offset level of the
ADCs (analog­digital converters) outside of
ARNICA's cryostat that was performed on
11mar95.
7

Table 4
Bad Pixel Counts
Date Number Number w/o 128,256 a Count levels, comments
1993 Jun 580 ---
1995 Jan 05 599 598 5900, 7800; low S/N.
1995 Nov 29 471 470 1900, 4000; vignetting left side.
1996 Nov 20 931 703 1500, 2500, 3500
1997 Sep 24 793 740 3000
1998 Jan 24 1203 764 700, 1400, 2050
a Not considering the bad columns 128, 256.
Second, the fraction of pixels in an ear­
lier mask contained in a later one varies
between 85 and 95%. This implies either
that the procedure to glean bp's from di­
rect images is only 85--95% reliable, or that
5--15% of the pixels defined as ``bad''are
only sporadically so, and depend on the
conditions of the electronics in that mo­
ment. We would argue that the latter ef­
fect is dominant because: 1) the number of
bp's gleaned from different images with the
same count levels acquired on the same day
are the same to within one or two pixels
(! 0.3%); and 2) different bad pixel selec­
tion algorithms detect similar sporadic be­
havior. There is also clear evidence for an
electronic origin of some bad pixels, as men­
tioned in the previous paragraph.
That some bp's can be sporadic is cor­
roborated by the finding that the number
of bp's gleaned from a given image depends
on count level, an effect also noted in early
1992­1993 data. The change in number of
bp's between count levels of 2000 and 4000
is roughly 25--30 pixels; this change was ver­
ified on two different dates for which we had
more than one image available. For count
levels between 700 and 1400, the number
changes by 150 or so pixels, although we
have only one date with these low flux
levels. That the behavior of ARNICA's
electronics influences the bp's is also sug­
gested by the sporadic presence of the ``bad
columns'' 128, 256; when they appear, they
seem to do so independently of flux levels,
and independently of any known effect.
Finally, the most recent images used to
derive the bp mask show a histogram (of
the original and smoothed image subtrac­
tion) with two peaks: a main peak around
zero as expected, and a secondary peak
towards negative values. The position of
the secondary peak, and the separation and
widths of the two maxima depend on the
count level. The secondary peak is centered
roughly on the negative value of the raw im­
age median (\Sigma 50 ADU), i.e., the original
value of these bp's was approximately zero.
The separation of the two peaks gets larger
with count level, as does their width. We
have some (weak) evidence that also this
behavior is due to the DC level prior to
the entrance of the ADCs: half of the ar­
ray appears to contribute more strongly to
the negative peak than the other half.
We conclude therefore that the behavior
of ARNICA's bad pixels is due partly to
real degradation of the cosmetic quality of
the array, and partly to variations in the be­
havior of the electronics. If we believe the
reproducibility of the bad pixel detection
algorithm, then we can say that these ``spo­
radic'' bad pixels comprise roughly 10% of
any given bad pixel list.
4. Concluding remarks
In the five or so years that ARNICA
has been in operation, it has proved to be
a rugged ``workhorse'', able to function ef­
ficiently in very different observing situa­
tions. This was made possible by ARNI­
CA's modular design; by S. Gennari and F.
8

Lisi who designed and implemented the var­
ious optical modules to match the different
telescopes (Gennari 1997); by C. Baffa, V.
Gavryusev, and E. Giani who adapted AR­
NICA's data acquisition software to differ­
ent telescopes and operating environments;
and by the organizational and observational
efforts of R.M. Stanga and L. Vanzi. We
hope that the future five years will be as
fruitful with the new NIR instruments as
these past five years have been.
REFERENCES
Gennari, S. 1997, Arcetri Observatory
Technical Report N. 2/97
Hunt, L.K., Calamai, G., & Oliva, E.,
1987, Arcetri Observatory Technical Re­
port N. 5/87
Hunt, L., Maiolino, R., & Moriondo, G.,
1994a, Arcetri Observatory Technical
Report N. 2/94 (Report I)
Hunt, L.K., Maiolino, R., Moriondo, G. &
Testi, L. 1994b, Arcetri Technical Re­
port, N. 3/1994 (Report II)
Hunt, L.K., Testi, L., Borelli, S.,
Maiolino, R., & Moriondo, G. 1994b,
Arcetri Technical Report, N. 4/1994 (Re­
port III)
Hunt, L.K., Lisi, F., Testi, L., Baffa, C.,
Borelli, S., Maiolino, R., Moriondo, G.,
& Stanga, R.M. 1996, A&ASS, 115, 181
Hunt, L.K., Mannucci, F., Testi, L.,
Migliorini, S., Stanga, R.M., Baffa, C.,
Lisi, F., & Vanzi, L. 1998, AJ, in press
Lisi, F., Baffa, C., & Hunt, L. 1993,
``ARNICA: the Arcetri Observatory
NICMOS­3 imaging camera'', Proceed­
ings of SPIE Technical Conference 1946,
Orlando
Lisi, F., Baffa, C., Biliotti, V., Bonaccini,
D., Del Vecchio, C., Gennari, S., Hunt,
L., Marcucci, G., & Stanga, R.M. 1996,
PASP, 108, 364
McCaughrean, M. J. 1988, Ph.D. Thesis,
University of Edinburgh
Oliva, E., & Origlia, L. 1992, A&A, 254,
466
This 2­column preprint was prepared with the AAS
L A T E X macros v4.0.
9