OPTICAL GRAVITATIONAL LENSING EXPERIMENT

                         ( OGLE )

                   EARLY  WARNING  SYSTEM

The "Early Warning System" (EWS) is a new feature implemented in the OGLE 
data system.  It allows the detection and identification of microlensing 
events in real time.  Thus, the system makes it possible to reschedule 
observations to obtain the best coverage of a microlensing event while it 
unfolds, as well as to arrange follow-up photometric and/or spectroscopic 
observations from other observatories.  The system was tested in April 1994 
during the first subrun of OGLE 1994 season, and also on the archive data 
from 1992 and 1993.

Currently OGLE monitors about two milion non-variable stars in the
Galactic Bulge every night with the Swope 1 meter telescope at the Las
Campanas Observatory.  These are stars that have shown no brightness
variations exceeding photometric errors over a period of two seasons
(1992 and 1993, about 1,100,000 stars in 13 fields) or one season (1993
only, about 900,000 stars in 7 fields).  The fields are identified in
two papers: Udalski et al, 1992, Acta Astronomica, 42, 253, and (1994),
44, 1.  Our 1994 observing season is from April 21 to September 15 in the
following subruns:

  Apr 21 - Apr 26
  Jun  7 - Jun 21
  Jun 30 - Jul 16
  Jul 24 - Aug  6
  Aug 19 - Aug 28
  Sep  5 - Sep 15


All new CCD frames are reduced within 24 hours, and the current magnitude 
of a non-variable object is compared with the mean magnitude from previous 
season(s).  If the object is brighter then a pre-assigned threshold a flag
is set in the EWS database.  When 5 sequential observations over the 
threshold are detected the EWS sends an alarm message to the operator. 

Following the EWS alarm message the following procedure is performed.  First,
the full light curve from the current season is inspected to check whether 
the candidate star behaved in a reasonable way over the entire season: are
the light variations smooth and do they resemble the rising branch of a
microlensing light curve.  Second, the most recent frame is inspected to
check whether the star is located in a good frame region, its measurements
are not contaminated by chip defects etc, and its brightness really increases.
Third, for at least three further nights the candidate is observed twice per
night and if the smooth brightness increase continues the candidate is marked
as the "prime microlensing candidate", and the message is e-mailed to the
interested observers.  It includes all the available photometric data for
the star, its coordinates, and a finding chart in a form of a PostScript 
subframe around the candidate: 150 x 150 pixels of 0.44 arcsec (66" x 66").

* What is the probablility that the "prime microlensing candidate" is indeed
  a microlensing event?

The EWS system as described was tested extensively on the archive
observing data from 1993 and 1992 seasons.  All microlensing events
reported to date by the OGLE (Udalski et al. 1993, Acta Astron., 43,
289; 1994, ApJ, 426, L69) were easily detected and identified in
"real-time", while no other object met the "prime microlensing
candidate" criteria.  For 1994 observing season the non-variable object
database is even more secure as for 13 out of 20 fields the database of
non-variable stars is based on two observing seasons.  The tests
performed in the real envinronment in April 1994 showed no microlensing 
candidates and no spurious alarms.  Thus, we estimate that the "hit
rate"  should be above 90%.  However, we cannot exclude the possibility
that some "prime microlensing candidates" might turn out to be
non-microlensing events as the selection is made on the basis of the
beginning of the rising branch only.  Nevertheless, any object
increasing its brightness after remaining constant for up to two years
is interesting and worth observing, anyway.   On the basis of our
1992/1993 experience we expect that between 2 and 4 events with a
characteristic time scale in excess of 10 days will be found by the OGLE
in 1994.

* What kind of follow-up observations are important?

Any photometric observations will improve light curve coverage of the event.
Photometric multicolor observations of the candidate are important to check
achromaticity of the event.  One should, however, be aware that the Galactic
Bulge fields are very crowded so a good seeing site is necessary to obtain 
precise photometry.  Spectroscopic observations may also help in establishing
the microlensing nature of the event and help to identify the lensing object
when the event is over.  The most important are frequent observations near
the maximum, as they may reveal the finite size of the source (Nemiroff and
Wickramasinghe, 1994, ApJ, 424, L21) or the presence of planets around the 
lens (Mao and Paczynski, 1991, ApJ, 374, L37).  Unfortunately, it is not 
possible to predict the time of maximum light with the early rising branch 
of the light curve on which the "prime microlensing candidate" is based, hence
catching the peak intensity is possible only by means of frequent observations.

* Referencing

Information about the on-going microlensing events will be distributed to
all interested observers via e-mail.  This information will be placed in 
"public domain" by making it acessible over anonymous ftp from

host:         sirius.astrouw.edu.pl (148.81.8.1)
directory:    /ogle
file:         ogle94.ews

Please feel free to redistribute it to others, but reference the source
of information, i.e. the OGLE project. It is expected that the OGLE
project will be acknowledged in any publication that will result from
using the OGLE EWS information.  

If you want to be on the OGLE EWS e-mailing list, please send your request to 

                udalski@sirius.astrouw.edu.pl

and describe briefly what kind of follow-up observations you intend to
make (please specify the site(s), instrument(s), expected ability to get
(use) the observing time). If you would like to accept a larger risk of
a false alarm and prefer to  receive the information at the time the EWS
sends its alarm to the operator (i.e. 3 nights earlier than standard) -
please state so clearly.  Please note, however that in this case a
fraction of false alarms may be substantial.


Andrzej Udalski and Michal Szymanski

Warsaw University Observatory
Warsaw, Poland