Cambridge, MA — Last year, astronomers at Wesleyan University announced
that they had discovered a “winking” star — a star known as KH 15D that
undergoes a regular, long-lasting (~20 day) eclipse every 48 days. They
theorized that those eclipses were caused by intervening blobs of material
within a protoplanetary disk surrounding that young star.

Spurred on by those findings, Harvard astronomer Joshua Winn
(Harvard-Smithsonian Center for Astrophysics) and colleagues decided to
examine the past behavior of KH 15D using sky photographs taken during the
first half of the 20th century and stored in the Harvard archives. They
found that the winking star used to not wink. The nearly complete eclipses
seen today were not happening several decades ago, meaning that the eclipses
now seen are a recent phenomenon that began within the past few decades — a
remarkably short time by astronomical standards.

“There are very few cases where astronomers can see a significant change to
a star over a single human lifetime,” said Winn. “And if the eclipses are
caused by material in a protoplanetary disk, as suspected, then that would
give us the exciting opportunity to study planet formation on surprisingly
short time scales.”

A Valuable Archival Resource

To probe the eclipse history of KH 15D, Winn and colleagues contacted
Harvard Plate Stacks acting curator Alison Doane, who identified more than
60 glass photographic plates containing images of the appropriate region of
space. All of the plates were taken between 1913 and 1955, a time long
before the advent of today’s widely used charge-coupled device (CCD) cameras
and digital storage media. Harvard’s Photographic Plate Collection contains
a half-million plates spanning a century of research from the 1880s to 1989,
making it both the largest such archive in the world and an irreplaceable
resource for astronomers seeking to study time-varying celestial phenomena.

Winn said, “Using the Harvard Plate Stacks is like having a time machine.
After an exciting object like KH 15D is discovered, you can go into the
stacks and observe it as it was nearly 100 years ago.”

Winn and colleagues examined the plates identified by Doane to look for
evidence that the brightness of KH 15D changed over time. Specifically, they
hunted for plates where stars of similar brightness could be seen but KH 15D
was absent, indicating that the winking star had dimmed due to an eclipse.

A brighter star close to KH 15D, combined with the intrinsic faintness of
the winking star, made the measurements a challenge. However, the
astronomers were able to identify about 40 photographic plates on which they
could measure KH 15D with sufficient accuracy to detect a deep eclipse.

If the star’s eclipses took place in the past just as they do today, then
approximately 16 plates (40 percent of the total) would have shown a dim,
eclipsed star. Instead, the astronomers found that none of the plates
definitively showed an eclipse.

“Statistically, we showed that it’s extremely unlikely that the eclipses
were taking place in the early 20th century with anything like their present
characteristics. Either they were much shorter in duration, or not nearly as
complete, or, they were not happening at all,” said Krzysztof Stanek (CfA),
co-author of the paper announcing their findings.

Co-author Peter Garnavich (University of Notre Dame) added, “Our most recent
observations show that the length of the eclipse is evolving rapidly. In a
few years, this strange star will spend more time faint than bright.”

Clues To Planet Formation

The star KH 15D is very much like our Sun, except that it is only a few
million years old versus the Sun’s age of 4.6 billion years. Star formation
theories predict that KH 15D may still be surrounded by a disk of dust and
gas left over from its birth. That disk, known as a protoplanetary disk, is
a possible source of the eclipses.

A nearly 3-week eclipse is difficult to explain by invoking an intervening
planet or companion star due to the length of the eclipse — the star’s face
is totally hidden from our view almost half the time. The most plausible
cause is a wide swath of disk material sliding in front of the star, thereby
blocking most of the star’s light.

One possibility is that this swath is actually a “ripple” in the
protoplanetary disk, recently stirred up by the gravitational influence of
an embedded protoplanet. A Jupiter-sized protoplanet orbiting some 0.2
astronomical units from the star could create such a ripple. (An
astronomical unit is the average distance between the Earth and Sun.)
Moreover, the ripple would evolve on a timescale of 10 to 100 years. This
makes protoplanet/disk interactions an appealing explanation for the
existence and evolution of the KH 15D eclipses.

“Still, there is no clear theoretical explanation for the eclipses,” said
co-author Dimitar Sasselov (CfA). “Radial velocity measurements can rule out
the intervening high-mass companion that has been suggested by some
researchers. That will leave us with two possibilities — either the
eclipses are caused by a ripple in a protoplanetary disk, or they are caused
by something we haven’t even thought of yet!”

Winn and colleagues now plan to collaborate with additional astronomers to
investigate other plate archives for data from the second half of the 20th
century. By studying when and how the eclipses began, they hope to gather
additional clues to their cause.

This research will be published in the August 20, 2003 issue of The
Astrophysical Journal Letters.

Headquartered in Cambridge, Massachusetts, the Harvard-Smithsonian Center
for Astrophysics (CfA) is a joint collaboration between the Smithsonian
Astrophysical Observatory and the Harvard College Observatory. CfA
scientists organized into six research divisions study the origin,
evolution, and ultimate fate of the universe.