These Level 2 data are made of calibrated velocities. Mainly due to the loss of the rotating 
mechanisms of the polarisors, the calibration is not trivial. Individual GOLF Team scientists 
developed their own technique, based on different assumptions.

2018-GOLF-22y Directory
New series calibrated as explained in Appourchaux, et al. (2018) (A&A 617 A108), containing: 
PM1, PM2, mean of the two signals.

2017-GOLF-Fossat_Series Directory
The time series used in the paper Fossat, et al (2017) (A&A, 604, id.A40). It is a 16.5-year 
series calibrated in velocities as explained in García et al. (2005) (A&A, 442, 385), that 
starts at 0:00:30 (T.A.I.) on April 11th 1996. It is sampled at 80 s.

2013-GOLF-17y Directory
One series covering data from April 11, 1996 to December 31, 2013. Data from both detectors 
(PM1 and PM2) are available, as well as the mean of the two PMs. We have reason to suspect 
a slightly enhanced noise contribution from PM1. The user can choose to avoid this by using 
only PM2, or to reduce counting noise by summing the two. The temporal cadence of the time 
series is of 60 seconds and the duty cycle is of 96. Please cite García et al. (2005) 
(A&A, 442, 385) when publishing results using these data.