The reduction of the negatives (e.g.
Whipple 1938; Ceplecha et al. 1979;
Ceplecha & Borovička 1992)
was performed at the Sterrewacht Leiden of the Leiden University and at
the Stichting Geavanceerde Metaalkunde of the Technical University of Twente.
Each set of star trails and
meteor trail were routinely measured in two directions to avoid hysteresis.
The meteor is measured twice per shutter break at the head of the break to
minimize the influence of wakes and trains.
The measurement accuracy is limited by the quality of the meteor and star trail
images, not by the measuring device, a Zeiss Astrorecord *X*-*Y*
measuring table,
with a nominal accuracy of 0.001 mm. The image quality is determined by
the quality of
the camera optics, the focal length, the brightness of the meteor, the
location on the negative, and by background fogging.
The standard deviation of measured star positions and the calculated grid is
about 30''-1' for the well focused F/2,
cameras with 35 mm film format
and the 60 mm F/4,
cameras that were
used in the small camera network. The all sky cameras, on the other hand,
use 35 mm cameras with
F/5.6, or F/2.8,
fish-eye
lenses, which results in a positional accuracy of about 3'-5'.
The meteor cameras are not guided but the instant of the beginning and ending of
each exposure is known with an accuracy of better than seconds and the
time of the meteor is usually known to seconds,
which allows for a
correction in RA of the stellar trails to the instant of the meteor to an accuracy
of about seconds (1.5'). Larger errors
occur when the time of the
meteor is incorrect due to a wrong identification
from the visual observations.
Large differences in time (more than a few minutes)
do usually not result in a reasonable solution of the
multistation calculation.
However, extra effort is made to ensure that the correct
identification is made.

The method of Turner (1907) is followed to calculate the stellar coordinates of the meteor path from the position of stars on the negative (Tadeusz 1983). The coordinates of the stars are taken from the Tirion Star Atlas 2000.0 (Sky Publishing), with 0.1'' accuracy. The method uses some 20 stars spread over the negative with a triangle of stars around the approximate plate center and adopts a coordinate grid that accounts for projection distortions from the optical system. This distortion becomes too strong for the method in the case of fish-eye images. Hence, for fish-eye images we used the method of Ceplecha outlined in the REDCON software routine (Ceplecha et al. 1979). This method makes use of seven independent constants to link the measured rectangular coordinates with the horizontal coordinates but demands that the center of the plate is near the zenith. If the number of usable stars for determining the coordinate grid is insufficient, simpler procedures with less constants are applied but with, of course, less precision.

The trajectory of the meteor in the atmosphere is calculated, subsequently,
by fitting planes through the station and meteor path.
The location of the stations and their altitude is known within 30 meters
or better.
The best plane is calculated by a least squares fit through the measured
positions of the meteor path. The combination of two planes then result in a
(partial) meteor path.
The mean trajectory is computed from all
individual paths with weight factors:

with *Q* the convergence angle and *W*_{i} = 1.0 for most images and
*W*_{i}= 0.7
for some less accurate fish eye components.

When a meteor is recorded by more than two stations, accuracy checks on
trajectory and radiant data can be made by comparing two-station solutions.
In most cases errors are caused by timing
errors, either in the exposure times or in the time of the meteor.
These errors are reflected in the scatter
in the Right Ascension (RA)
of the position of the radiant in the various solutions, but do also affect
the Declination. Small convergence angles between the planes
affect the radiant and the geographic data of the trajectory in a negative way.
It is found that for convergence angles (*Q*) larger
than 40 degree, the
error in radiant position is about . For *Q* = 20 degree
it is
about and for *Q* = 10 degree
it is about . If the
convergence angle is less than 10 degree, then the accuracy is not better
than radio orbits of order
(Sekanina 1976).
As much as 41 of the 359 precisely
reduced meteor orbits in our study do not lead to precise results,
because they have , but are still considered valuable in some
cases due to the small number of known stream orbits, the large mass of the
meteoroid, or a relatively high accuracy in a part of the orbital elements.
Error estimates are given for each individual orbital element (Table 2 (click here)).

Code - DMS sequential numbering starting with the year |

Month - month |

Dec. Day - day and time (UT) in decimal days |

N - number of multi-station components |

Stream - meteor stream identification |

M_{v} - absolute visual magnitude |

q - perihelion distance (AU) |

a - semi major axis (AU) |

e - eccentricity |

i - inclination (Eq. 2000) |

omega - - argument of perihelion (Eq. 2000) |

Node - - ascending node (Eq. 2000) |

pi - - longitude of perihelion (Eq. 2000) |

- geocentric velocity (km/s) |

- heliocentric velocity (km/s) |

- apparent pre-atmospheric velocity (km/s) |

V - average velocity along trajectory |

- beginning height (km) |

- height of brightest point on meteor track (km) |

- end height of meteor (km) |

RA obs. - apparent right ascension of radiant (2000) |

[+/-] - error due to uncertainty in time of meteor |

DEC obs. - apparent declination of radiant (2000) |

RA Geo - geocentric right ascension of radiant (2000) |

Dec Geo - geocentric declination (2000) |

CosZR - cosine of zenith angle of radiant at time of meteor |

- maximum convergence angle between planes. |

The initial velocity is computed by making use of a fit through all
computed positions of the meteor at the head of shutter breaks
(Jacchia & Whipple 1961):

with *L* being the geographic position
of the trail, *H* the height, while and
*K* are variables.
It is found that this fit gives reliable results only for meteor trails with
at least some 30 well measured breaks.
In cases with less breaks (*N* = 15-30) a fit of the following equation
is used instead
(Ceplecha & Borovička 1992):

where is the air density at a
given height (or time *t* of the meteor)
computed from the CIRA 1961 reference atmosphere
(Kallmann-Bijl et al. 1961).
The final accuracy with which is determined is about 1% in case
a sufficient number of breaks can be measured. The correction from
measured velocity at the beginning of the trajectory to the pre-entry
velocity before deceleration in the atmosphere
varies from 0.2 km/s for fast meteors up to 1.5 km/s for
some slow meteors.

Given an accurate radiant position, the uncertainty in the orbit is mainly determined by the uncertainty in the velocity determination. That demands stable rotating shutters. Before 1986, all our rotating shutters relied on the stability of the mains frequency, which is thought to be constant and accurate at 50 Hz to within 0.5%. The chopping frequency of the shutters was measured to be accurate within 0.2%. However, the rotational period of the shutters can oscillate with a variable oscillation period and an amplitude up to 4%. Such oscillations occur when the rotating shutters are perturbed by strong winds or when there are system resonances. Such oscillations occur at random. Larger than 1% errors are seldomly found when information from more than one rotating shutter is available. To improve the quality of the computed orbits, the Dutch Meteor Society introduced crystal controlled rotating shutters in 1986 at some stations. Starting in 1992, all stations were equipped with new crystal controlled rotating shutters with twice the shutter frequency, increasing the shutter speed to 50 breaks per second with a stability of 0.1% or better.

The orbital elements (J2000) follow from the true radiant position, the computed initial apparent velocity and the position of the Earth at the time of the meteor (e.g. Katasev 1957). The calculations take into account such corrections as zenith attraction, the curvature of the Earth, and diurnal aberration, and give good results even at shallow entry angles and large distances between the stations (Ceplecha et al. 1979; Ceplecha & Borovička 1992). That was demonstrated again in one case DMS 85027, a sporadic fireball, which was photographed by four Dutch stations as well as by four distant German stations of the European Network (Betlem & de Lignie 1985). Independent reduction in Leiden and Ondřejov gave good agreement between the Dutch and German stations (Ceplecha, private communication).