2 The data

Our data set consists of observations from two separate campaigns, namely the VLBA monitoring of EGRET-detected blazars by Jorstad et al. (2001a) and the Mets $\ddot{\rm a}$hovi Radio Observatory quasar monitoring program. Jorstad et al. (2001a) monitored a sample of 42 $\gamma$-bright blazars at 22 and 43 GHz with VLBA between 1993 and 1997. For 27 of these sources (see Table 1), there were enough TFD variation data available from Mets $\ddot{\rm a}$hovi monitoring (Ter $\ddot{\rm a}$sranta et al. 1998) to reliably identify large outbursts in the flux curves.

Selection criteria for sources observed by Jorstad et al. (2001a) with the VLBA were: (1) detection by EGRET (E > 100 MeV; Hartman et al. 1999); (2) flux density at 37 GHz $\gtrsim$ 1 Jy; and (3) declination (J2000) $\ge -30 \hbox{$^\circ$ }$. These criteria give a sample containing roughly 60% of the known $\gamma$-ray blazars (Hartman et al. 1999). The VLBA observations were made at high radio frequencies giving $\approx$0.1-0.3 mas resolution and $\approx$10 mas map size. High resolution allows us to study the inner parts of the jet and possibly see how the shock formation is connected to the flaring behaviour of these sources. However, as we discuss below, VLBA maps at 43 GHz in many cases still have insufficient resolution to separate the new shock in the jet from the core before the millimetre flare is over.

Since its beginning in 1980, the Mets $\ddot{\rm a}$hovi quasar monitoring program has been the most comprehensive such program at high radio frequencies. The Mets $\ddot{\rm a}$hovi sample contains 157 individual sources including about 100 of the brightest radio-loud AGN in the Northern hemisphere (declination $\gtrsim -10 \hbox{$^\circ$ }$), which are observed at 22, 37 and 87 GHz (see Ter $\ddot{\rm a}$sranta et al. 1998 for details). The sample also includes the Northern 2 Jy catalogue of flat spectrum sources (Valtaoja et al. 1992a) fulfilling the following criteria: $\delta \ge 0 \hbox{$^\circ$ }$, $\alpha _{{\rm Kuhr}} (2.7$- $5~\textrm{GHz}) \ge -0.5 \quad (S \propto \nu ^{\alpha})$, with $\alpha$ taken from the catalogue of Kühr et al. (1981), and $S_{{\rm max}}(22 \;\mbox{GHz}) \ge 2$ Jy. Of the 27 sources in our study, 13 belong to this 2 Jy catalogue. Of the 14 sources that do not belong, 5 have declination below $0 \hbox{$^\circ$ }$ and the rest were fainter than 2 Jy at 22 GHz prior to 1992. In our study, we used 22 and 37 GHz Mets $\ddot{\rm a}$hovi data from 1990 to 1998 together with 22 and 43 GHz VLBA maps. Comparing 37 GHz TFD-data with 43 GHz VLBA maps is justified by the typically flat spectra of our sources in the millimetre region.

Our sample of 27 sources with good VLBI and TFD data consists of 12 high optical polarization quasars (HPQs), 7 low optical polarization quasars (LPQs), 7 BL Lacertae objects (BLOs) and one object classified as a radio galaxy (GAL). The percentage of each class of radio-loud AGN is given in Table 2 for our sample, for the 2 Jy sample (Valtaoja et al. 1992a), and for a sample containing the EGRET blazar identifications that have a high probability of being correct (Mattox et al. 2001). As one can see, our sample is very similar to the $\gamma$-ray blazars as well as to the 2 Jy sample representing the brightest radio-loud AGN. [The one radio galaxy in our sample, 0446+112, is a less certain EGRET identification, which is the reason why it is not included in the list by Mattox et al. (2001).] This supports the notion that the results presented in this paper are applicable to all $\gamma$-ray blazars and, to some extent, to radio-loud AGN in general.