3 Results

Table2 contains the basic source information for the 1990 April and 1993 April sessions for sources that yielded fringes at either or both epochs. The table gives the positions in the J2000.0 coordinate system, the cosmological red-shift, the optical magnitude, the source flux density inferred from the shortest baselines (Flux$_{\rm VLBI}$), and the measured single dish flux density (Flux$_{\rm SD}$) at $\lambda$3mm (Tornikoski et al. 1996) where this information is available for the source in question. Table3 contains more detailed source information. It gives the maximum observed proper motion in the source (Vermeulen & Cohen, 1994 and references therein), measured X-ray (Della Ceca et al. 1990), and maximum observed $\gamma$-ray fluxes for photon energies >100 MeV (Chiang et al. 1995; Lin et al. 1993; Montigny et al. 1995).

  

Table 2: Basic source information for the April 90 and April 93 sessions for sources that yielded fringes. Positions are in the J2000.0 coordinate system, z is the cosmological redshift, m_v is the mean visual magnitude, Flux$_{\rm VLBI}$ is the source flux density inferred from the shortest (continental) baselines, and Flux$_{\rm SD}$ is the measured single dish flux density at $\lambda$3mm at the time of the observations. Where available the information has been given for both the 90 and 93 epochs. Positions, redshift, morphological type, and m_v are taken from the NED database

  

Table 3: Source information. $\mu$, is the maximum proper motion observed in this source at lower frequencies. $F\rm _X$ is the measured X-ray flux, and $F_\gamma$ is the maximum observed $\gamma$-ray flux for photon energies >100MeV

For all mapped sources we present the maps with a 50$\mu$as circular beam if the original beam is within 10% of this value. The size of the convolving beam was chosen to be able to compare with the previously published maps (Bååth et al. 1992). Table4 gives the UV data information. It gives the number of scans with closure phase data and the length of time over which the sources was observed for the epochs in question.

   Table 4: UV data information. Scans, are the number of scans with closure phase information and Time, is the length of time over which the scans where found. The 1990 epoch used an observing scheme where the data were recorded for 6.5 minutes and followed by a 23.5 minutes idle time giving 30 minutes for one observation cycle. With more tapes available for the 1993 epoch, the idle time was only 6.5 minutes giving a total observation cycle of 15 minutes

3.1 3C84

The radio source is identified with the dominant disturbed elliptical galaxy in the Perseus cluster. It has been detected as a strong diffuse X-ray source with a strong X-ray point source coincident with the radio core. The radio frequency morphology is quite complicated (Krichbaum et al. 1992; Venturi et al. 1993) exhibiting a core with two opposite radio-jets, with the southern jet consisting of components moving down a diffuse jet and finally expanding into an amorphous component at 12mas. Krichbaum et al. (1992) showed that the inner jet components move with 0.1c and that after a major bend the jet speed accelerated. $\lambda$7mm VLBI observations (Krichbaum et al. 1993) show a complex structure with 6 components embedded in an extended jet.

  

Figure 1: 3C84 100GHz map from 1990 April 21. Peak flux density = 1.1 Jy/Beam. Contour intervals are chosen as (-2.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0,128.0) $\times$ 10 mJy/beam. In this and succeeding figures, negative contours, if any, are shown dashed. The restoring beam was a elliptical Gaussian (0.12 $\times$ 0.03 mas in PA $=-21.3\hbox{$^\circ$}$)

The 3mm (US-only), epoch march 1987 map by Wright et al. (1988) exhibits a core with a jet in PA $\sim$ 230$\hbox{$^\circ$}$. Embedded in the jet at $r\sim 0.3$mas is a second component. The previous $\lambda$3mm observations (Bååth et al. 1992) showed a core with a component moving out with $\mu =85$ $\pm$ 7$\mu$as/year with diffuse components 0.5mas south of the core. In the 1989 map there is a hint of a ridge-line connecting the components but the dynamic range is not sufficient to clearly show the underlying flow.

  

Figure 2: 3C 84 86GHz map from April 4, 1993. Peak flux density = 0.74 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a elliptical Gaussian (0.07 $\times$ 0.05 mas in PA $=-68.7\hbox{$^\circ$}$)

The 1990 epoch map (Fig. 1) has a structure very similar to that seen at the previous epochs. A core extended in PA $\sim -90\hbox{$^\circ$}$(A) with two components well separated from the core, one in PA $\sim 200\hbox{$^\circ$}$ at $r\sim$150$\mu$as(B) and the other in PA $\sim 180\hbox{$^\circ$}$ at $r\sim 0.7$mas(C). This structure agrees very well with that seen at the previously published epochs (Wright et al. 1988; Bååth et al. 1992). The 1993 map (Fig. 2) has a unresolved core(A) and two components in PA $\sim 180\hbox{$^\circ$}$, one extended component at $r\sim 0.35$mas(D) and a weak component at $r\sim 0.8$mas(E). With 3 years between the epochs it is not surprising that the structure has changed dramatically. Looking at the three epochs from 1989 to 1990 it appears as if the components leave the core in a PA $\sim -100\hbox{$^\circ$}$, and later turn to the south. In Table2 it is clear that most of the total flux density is resolved out, and that our $\lambda$3mm maps probe only the most compact structures, leaving the flux density and position of more extended emission ill constrained. This fact is valid for most of the sources presented in this paper unless specified otherwise. The structures in both maps were too complicated to allow a Gaussian model fit to the uv data, as the fitting program has severe problems fitting more than three components to a uv data set. Thus we present the result of fitting Gaussian components to the images (Table5). As the positions of the components are the result from fitting to the image we have also labelled the components in the two maps.

  

Table 5: Model parameters for Gaussian components fitted to the 3C84 maps for sessions 90 and 93. Flux is the flux density of the component, $\Delta\alpha$ and $\Delta\delta$ gives the position of the component relative to the core, Size is the size of the fitted Gaussian, PA is the Position Angle of the major axis of the fitted Gaussian. The errors represent 3$\sigma$ from the fit

It is clear from Figs. 1 and 2 that there are drastic internal structural changes in the source. The long time between epochs makes identification difficult, but we will make some simple assumptions to try to quantify these structural changes. If we assume that component D is component B seen at a later epoch and making the same assumption for components C and E, we get a proper motion of $70 \pm 20 \mu$as/year for component B and $40~\pm ~10~\mu$as/year for component C. These values agree well with the velocities seen by Bååth et al. (1992). If component B is the result of a component leaving the core after the March 1989, as suggested by the extension in PA $\sim -135\hbox{$^\circ$}$ in the March 1989 map (Bååth et al. 1992) and the observed increase in flux density, then we can estimate the proper motion to be $\sim 130 \pm 20$ $\mu$as/year.

3.2 3C111

This is a classical double-lobed FRII radio-source with an extended jet in PA $\sim-45\hbox{$^\circ$}$ (Preuss et al. 1990). Components are seen moving out from the core at superluminal velocities $\sim$1mas/year.

  

Figure 3: 3C111 86GHz map from April 4, 1993. Peak flux density = 0.8 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

In Fig. 3 we present the highest resolution map of this source available to this date. The core is unresolved and there is a weak component in the same PA as the mas scale jet at a distance of 0.5mas. The result from model fitting Gaussian components to the UV data can be found in Table6.

  

Table 6: Model parameters for the Gaussian components fitted to the 3C111 uv data. Explanation of parameters can be found in Table5

The Gaussian model fit clearly substantiates the general structure presented in the hybrid map. With only one epoch we are unable to determine whether the superluminal motion seen at lower frequencies (Preuss et al. 1990) is present on $\mu$as scales. Further observations are needed to detect the possible proper motion of components in this source.

3.3 0420-014

This source is a radio-loud flat-spectrum AGN and also classified as a Blazar. It has been detected in $\gamma$ and X-rays (Radecke et al. 1995), and has a very high optical polarization of $\sim$17% (Wills et al. 1992). At mas resolution the source shows a symmetrical unresolved core. This is the case of either a "naked" core or that of a jet aligned very close to the line of sight (Wehrle et al. 1992). At $\lambda$7mm the source consists of a core and a pronounced bent jet (Krichbaum et al. 1994a). We found fringes to the source in 1990 for one scan which was not enough to make a hybrid map but sufficient to make a fit to the UV data. The result of the model fit is presented in Table7.

  

Table 7: Model parameters for Gaussian components fitted to the 0420-014 UV data. Explanation of parameters can be found in Table5

Comparing with the Flux$_{\rm SD}$ (Table2), it is clear that more than 50% of the flux density is missing. With so few uv points covering such a short time it is difficult to make any statements about the source structure, but it is clear that this source is observable at $\lambda$3mm, and warrants further observations.

3.4 0735+178

This source is a BLLac object which is point-like in the optical and with arcsec resolution at radio wavelengths. VLBI polarization observations at 6cm show that the core has a high degree of polarization, 3% (Gabuzda et al. 1994). At mas resolution the source has an unresolved core and a jet extending to the NE (Bååth & Zhang 1991), with components moving at superluminal velocities. The components appear to follow different paths as they move out from the core (Gabuzda et al. 1994).

  

Figure 4: 0735+178 100GHz map from April 20, 1990. Peak flux density = 1.2 Jy/Beam. Contour intervals are chosen as (-2.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0) $\times$ 20 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Our April 90 map, the first ever of this source with $\mu$as resolution, shows (Fig. 4) two components but it is not clear which is the core. There are no significant features outside the region shown. The result from model fitting Gaussian components to the UV data can be found in Table8.

  

Table 8: Model parameters for Gaussian components fitted to the 0735+178 UV data. Explanation of the parameters can be found in Table5

As with 3C111 we have a good agreement between the hybrid map and the model fitting. Almost all of the Flux$_{\rm SD}$(Table2) is seen in our map suggesting that most if not all high frequency flux density is emitted from this central region. Having only one epoch available we can say nothing about the superluminal motion in this source at $\mu$as scales. Assuming that component B is the "core'' the jet is straight from $\mu$as to mas scales; if component A is the "core'', this would indicate a sharp twisting of the jet from $\mu$as to mas scales. The identity of the "core'' is currently unknown.

3.5 0748+126

This is an unclassified QSO whose redshift is ambigous; two possible redshifts are suggested by Wills & Wills (1976): z=0.281 or z=0.88. VLA observations (Murphy et al. 1993) show an unresolved core with several components in PA $\sim 135\hbox{$^\circ$}$ connected by a slightly bent jet.

  

Figure 5: 0748+126 86GHz map from April 4, 1993. Peak flux density = 0.8 Jy/Beam. Contour intervals are chosen as (-2.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

The first map of this source made with $\lambda$3mm VLBI is presented in Fig. 5. The map also shows an unresolved core with a component in PA $\sim 90\hbox{$^\circ$}$ at 0.37mas. The weak components do not lie in the general PA of the VLA map, but as in many other sources this is not an uncommon feature (see 3C345, 3C446, OJ287, BLLac, and CTA102 in this paper). With no single dish monitoring available for this source at this epoch we cannot make any conclusions concerning the degree of missing flux density. To test the fidelity of the hybrid map we also made a Gaussian model fit to the UV data. The best fit is presented in Table9, which confirms the flux densities and location of components seen in the hybrid map.

  

Table 9: Model parameters for Gaussian components fitted to the 0748 + 126 UV data. Explanation of the parameters can be found in Table5

3.6 OJ287

This is a prominent member of the BLLac class with highly varying flux densities in both the optical and radio regimes. The variability time-scales vary from minutes to years (Takalo 1994; Aller et al. 1985). The source exhibits superluminal expansion along PA $\sim -110\hbox{$^\circ$}$ (Roberts et al. 1987; Gabuzda et al. 1989). Model-fitting to the 7mm VLBI data (Krichbaum et al. 1993) yields a two component model with the second component at 0.9mas in PA $=232\hbox{$^\circ$}$. The previous two epochs (Bååth et al. 1992) show an unresolved core.

  

Figure 6: OJ287 100GHz map from April 21, 1990. Peak flux density = 1.7 Jy/Beam. Contour intervals are chosen as (-2.0, 2.0, 4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

The April 1990 map (Fig. 6) shows an unresolved core. The uv data for this epoch has been discussed extensively by Tateyama et al. (1996). Here we include the map for completeness. In 1993 (Fig. 7) the core is still unresolved and there is a component in PA $\sim-45\hbox{$^\circ$}$ at 0.05mas.

  

Figure 7: OJ287 86GHz map from 1993 April 4. Peak flux density = 0.9 Jy/Beam. Contour intervals are chosen as (-1.0, 1.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Most of the single dish flux density (Table2) is seen in our map, the missing flux density may be the result of either incorrect calibration and/or resolution effects. In the three years between the observations the Flux$_{\rm VLBI}$ has decreased by 40%. We attribute the decrease in observed flux density to component B moving out from the core and decreasing in strength. The result from the Gaussian model fit to the uv data can be found in Table10.

  

Table 10: Model parameters for Gaussian components fitted to the OJ287 UV data. Explanation of the parameters can be found in Table5

There is a good agreement between the hybrid maps and the Gaussian model fits at both epochs. Assuming that component B has moved out from the core after the 1990 April epoch, we can obtain a lower limit to its proper motion. In the 1993 April map, component B lies at $r\sim 102 \pm 35\ \mu$as, thus it has moved with a minimum proper motion of 34 $\mu$as/year. Our $\lambda$3mm data suggests that components leave the core in PA $\sim-45\hbox{$^\circ$}$ and later turn to PA $\sim -110\hbox{$^\circ$}$ on mas scales.

3.7 1055+018

This source has a core-jet structure at mas resolution (Bondi et al. 1996) with a 15mas jet extended in PA $\sim -65$ and a weak component at $\sim$25mas in the same general PA. On arcsec scales the source has a triple structure, a strong central component and two diffuse lobes in a NS direction. The southern feature is connected to the central bright component with a bent jet (Murphy et al. 1993).

  

Figure 8: 1055+018 100GHz map from April 20, 1990. Peak flux density = 1.11 Jy/Beam. Contour intervals are chosen as (-1.0, 1.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Figure8 shows the first 50$\mu$as map of this source. The hybrid map shows an unresolved core with an extended jet in PA $\sim -135\hbox{$^\circ$}$. As for OJ287, half of the single dish flux density (Table2) is lost in the mmVLBI map, suggesting that the missing flux density is emitted from the jet which is resolved out in our map. To test the hybrid map we tried model fitting Gaussian components to the uv data. The best fit is presented in Table11.

  

Table 11: Model parameters for Gaussian components fitted to the 1055+018 UV data. Explanation of the parameters can be found in Table5

The hybrid map and the Gaussian model fit disagree in the exact location and the flux densities of the two components. Both models have two strong components separated by $\sim 150 - 200\,\mu$as. The Gaussian model fit places the second component more to S than in the map. We conclude that the sparse UV coverage makes the interpretation of the data very difficult. Observations with fuller uv-coverage are needed to determine its exact structure. For now we conclude that the data supports a structure with two strong components separated by $\sim$175$\mu$as.

3.8 3C273B

This quasar is a classical superluminal source (Zensus 1987). The mas jet can be seen extending from the core out to more than 150mas (Unwin 1990), and the PA of this jet is well aligned with the arc-second scale jet (Davis et al. 1985). VLBI monitoring of this source at $\lambda$1.3cm (Zensus et al. 1990) shows two features moving out from the core, with $\mu =0.65$ and $\mu =0.92$ mas/yr respectively. A prominent feature is the twisting of the jet, seen in both total intensity maps and polarization maps (Leppänen et al. 1995), with an increase in polarization with distance from the core.

A 3mm observation made by Krichbaum et al. (1990), showed jet components being ejected after a major optical outburst. Bååth et al. (1991) showed a core with a bent jet, with several components at different PA's, suggesting that the wiggling seen at mas scales continued at $\mu$as scales. The 1988 map showed an elongated component emerging at the time, which could be related to an outburst, seen 60 days earlier in Optical/IR (Courvoisier et al. 1988). Krichbaum et al. (1996b) showed 2 new epochs (1994 & 1995) which clearly show the fast (0.5-0.6 mas/yr) superluminal motion in 3C 273.

  

Figure 9: 3C273B 100GHz map from April 20, 1990. Peak flux density = 3.0 Jy/Beam. Contour intervals are chosen as (-8.0, 8.0, 16.0, 32.0, 64.0, 128.0, 256.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

The 1990 map (Fig. 9) shows an unresolved core and a component in the same PA as seen in earlier 3mm maps of this source. We do not see the other components but they may be too weak to be detected with the limited dynamic range we have in this map. Most of the single dish flux density (Table2) is missing, suggesting that the major part of the $\lambda$3mm flux density is emitted by the extended jet. The result from a Gaussian model fit to the UV data can be found in Table12. Both methods agree on the general location and flux densities of the fitted components.

  

Table 12: Model parameters for Gaussian components fitted to the 3C273B UV data. Explanation of the parameters can be found in Table5

We are unable to determine the proper motion in this source as the previous epoch map was made in 1998 March and the structural changes have been too large to identify the components and determine their motions.

3.9 3C279

This has been one of the most frequently observed sources since it was the first radio object to exhibit superluminal motion (Whitney et al. 1971). Cotton (1979) confirmed the superluminal motion with a series of observations showing a expansion velocity of 0.5mas/yr in PA $\sim -140$. Later observations showed lower expansion speeds $\sim$0.15mas/yr (Unwin et al. 1989; Carrara et al. 1993). The arcsec structure is a jet extending out to 10arcsec in PA $\sim -145\hbox{$^\circ$}$ (de Pater & Perley 1983); the jet is straight from mas to arcsec scale.

From being a moderately strong source in radio to $\gamma$-rays the source increased in flux density drastically such that in 1992 it was the brightest extra-galactic $\gamma$-ray source in the sky (Maraschi et al. 1994).

The previous epoch observations (Bååth et al. 1992) showed a strong core with two weak components. That paper gave an estimated proper motion (0.15 mas/yr) which agrees with that seen at much lower frequencies.

  

Figure 10: 3C279 100GHz map from April 21, 1990. Peak flux density = 2.0 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Our 100GHz map from April 1990 (Fig. 10) has a similar structure to that seen in 1989. With the improved UV-coverage we are able to map the source from $\mu$as to mas scales. The PA of the components agree very well with the PA seen at mas scales (Carrara et al. 1993). The 1993 epoch map (Fig. 11) shows an unresolved core with two strong components in PA $\sim -145\hbox{$^\circ$}$. The differences in position of the components at the two epochs clearly demonstrates structural changes in the source.

  

Figure 11: 3C279 86GHz map from April 4, 1993. Peak flux density = 1.1 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Comparing the Flux$_{\rm SD}$ with Flux$_{\rm VLBI}$(Table2) shows that the $\lambda$3mm emission of this source, like 3C273, is strongly dominated by the extended jet. As for 3C84 the structure is to complicated to make a Gaussians model fit to the UV data. The result of fitting Gaussian components to the images can be found in Table13. As the positions of the components are the result of fitting to the image we have also labelled the components in the two maps.

  

Table 13: Model parameters for Gaussian components fitted to the 1990 and 1993 maps of 3C279. Explanation of the parameters can be found in Table5

The large separation in time between the April 1990 and April 1993 epochs and the apparent rapid structural changes makes identification and determination of proper motion using only these two maps very difficult. Fortunately we have the March 1989 map which is only separated by a year from our 1990 map. Assuming that component B in the April 1990 map is the outermost component seen in the March 1989 map, labelled F2 (Bååth et al. 1992), we estimate the proper motion to be $\mu\sim 80 \pm 10~\mu$as/year. This is a lower value than previously seen at $\mu$as scales, but our value should be seen as a lower limit to the proper motion, as it is based on the conservative assumption that component B has moved from the position of F2 and not from the core, labelled F in the 1989 map.

3.10 1510-089

VLBI observations at $\lambda$18cm (Bondi et al. 1996) of this low frequency variable show a core-jet structure, with the jet extending 5mas in PA $\sim 180\hbox{$^\circ$}$. We found fringes to the source in 1990 for two scans, not enough to make a hybrid map, but sufficient to make a fit to the uv data. The result of the fit is presented in Table14.

  

Table 14: Model parameters for Gaussian components fitted to the 1510-089 UV data. Explanation of the parameters can be found in Table5

The resulting model fit has a Gaussian component elongated in NS, this elongation may very well be the result of poor uv-coverage and should not be interpreted as actual source structure.

3.11 3C345

The quasar 3C345 has been observed over a large range of wavelengths, from radio to hard X-rays. Most of the total power of 3 10$^{\rm 40}$W is emitted in the sub-millimeter-optical domain, with one-sixth of the power radiated at radio and X-ray wavelengths (Bregman et al. 1986; Ku et al. 1980). At wavelengths longer than $\lambda 3$mm, the radio continuum spectrum is flat, but steepens at wavelengths shorter than 3mm from a power-law of $\nu^{\rm{-0.91}}$ over the band $\lambda 3$mm $-~\lambda 30\,\mu$m, to $\nu^{\rm{-1.40}}$ over $\lambda 30$$\mu$m $-~\lambda 3\,\mu$m. 3C345 has provided one of the best cases for superluminal expansion and acceleration (Unwin et al. 1983; Moore et al. 1983). Observations made at several epochs and wavelengths show that the components seem to move outwards along a helically curved jet (Biretta et al. 1986; Steffen et al. 1993; Zensus et al. 1995). Different components appear to follow different paths as they move out from the core (Steffen et al. 1993; Zensus et al. 1995). 7 mm maps made at the time between our two epochs (Krichbaum et al. 1993) show a core with several components at different PA's. The PA changes from $-45\hbox{$^\circ$}$ at 0.2mas to $\sim -130 \hbox{$^\circ$}$ at 3mas. The 3mm maps (Bååth et al. 1992) show a central feature assumed to be the core and components leaving the core in PA $\sim-45\hbox{$^\circ$}$, and the jet changing PA to $\sim -135\hbox{$^\circ$}$ at 0.4mas, showing that the jet twists dramatically as it leaves the core.

  

Figure 12: 3C345 100GHz map from April 20, 1990. Peak flux density = 1.8 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

The source was detected in the 1990 session and shows (Fig. 12) an elongated core in PA $\sim -135\hbox{$^\circ$}$. Comparison between Flux$_{\rm SD}$ and Flux$_{\rm VLBI}$ (Table2) shows that this source like 1055+018 is largely resolved and that probably most of the missing flux density will be found on larger scales. The best model fit was achieved by using two Gaussian components. The actual location of the weaker component varied for model fits with similar rms, the location of the second component as presented in Table15 is an average of these model fits.

  

Table 15: Model parameters for Gaussian components fitted to the 3C345 UV data. Explanation of the parameters can be found in Table5

We conclude that both the image and the fits to the UV data confirms the existence of a second component in PA $\sim -90\hbox{$^\circ$}$although the exact position is not determined.

The poor quality of the April 1990 map makes determination of proper motions impossible. Component B could be component E3 from the April 1989 map, or a completely new component moving out from the core. With the available data we can only say that 3C345 exhibits drastic structural changes even on timescales as short as a year.

3.12 2145+067

Gregorini et al. (1984) showed that this object is a low frequency variable, which complicates the interpretation of the structure seen at lower frequencies. VLBI observations at 5GHz (Wehrle et al. 1992) show a core extended in PA $\sim 140\hbox{$^\circ$}$ with a weak diffuse component at $\sim$10mas in the same general PA. The core also appears to be embedded in a weak halo. VLA observations (Murphy et al. 1993) show a source with a second component $\sim$3arcsec away in PA $\sim-45\hbox{$^\circ$}$, thus the extension seen at mas resolution is almost 180$\hbox{$^\circ$}$ away. This suggests that either the low frequency variability may affect the perceived structure or the jet twists dramatically as it moves out from the source.

This source has been detected at 215 GHz with an SNR of 124 on the baseline Pico Veleta - Plateau de Bure (Greve et al. 1995). Even at this high frequency it is therefore very bright and compact.

  

Figure 13: 2145+067 86GHz map from April 4, 1993. Peak flux density = 1.05 Jy/Beam. Contour intervals are chosen as (-1.0, 1.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Our hybrid map shows a core weakly elongated in PA $\sim -120\hbox{$^\circ$}$. No significant structure is seen. The result of a Gaussian model fit to the uv data can be found in Table16.

  

Table 16: Model parameters for Gaussian components fitted to the 2145+067 UV data. Explanation of the parameters can be found in Table5

3.13 BLLac

BLLac exhibits very weak emission lines in the optical continuum above 5Å (Stickel et al. 1993; Vermeulen et al. 1995). It is strongly variable at all wave-bands (Angel & Stockman 1980; Aller et al. 1985), and shows dramatic variations in polarization (Gagen-Thorn et al. 1994). VLBI observations at 10.7GHz show a stationary core with components moving out with superluminal velocities in PA $\sim 190\hbox{$^\circ$}$ (Mutel et al. 1990). Recent observations at $\lambda$18cm (Bondi et al. 1996) show a single component extended in the NS direction. Previous maps by Bååth et al. (1992) showed an unresolved core with components moving out in PA $\sim -90\hbox{$^\circ$}$ with a proper motion $\ge$90$\mu$as/yr. The 3mm 1988 epoch map convolved with a 0.5mas circular Gaussian show a weak extended component at 1.5mas in PA $\sim 200\hbox{$^\circ$}$ agreeing with the structure seen at 10.7GHz.

  

Figure 14: BLLac 100GHz map from 1990 April 20. Peak flux density = 1.3 Jy/Beam. Contour intervals are chosen as (-4.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a 0.05 mas (FWHM) circular Gaussian

Our map from 1990 April (Fig. 14) shows a core-jet structure with the jet in slightly different PA compared to the previous epoch at 100GHz. This suggests that component E1 (Bååth et al. 1992) has moved in the general direction of the jet seen on larger scales. Most of the Flux$_{\rm SD}$ is seen on the shortest baselines (Table2) suggesting that the emitted radiation at $\lambda$3mm originates from the core. Some of the short baseline flux density has been lost in the mapping process. The result from model fitting Gaussian components to the uv data can be found in Table17. There is good agreement between the two methods. The extended jet seen in the image is very clear in the Gaussian model fit where it is represented by an elongated component.

  

Table 17: Model parameters for Gaussian components fitted to the BLLac UV data. Explanation of the parameters can be found in Table5

The 1989 March map has a component E1 at $r\sim 90~\mu$as in PA $\sim -90\hbox{$^\circ$}$. This component appears to have moved out and disappeared in our 1990 map. Assuming that component B has moved out from the core in the intervening time, then a lower limit to its proper motion is 63 $\pm$ 8 $\mu$as/year. This is a lower value than previously seen in this source (Bååth et al. 1992), but it is only a lower limit and may well be much higher.

3.14 3C446

The 100GHz map on 3C446 for the 1990 April epoch has been discussed extensively in Lerner et al. (1993). The map shows a central core with a component in PA $\sim -140\hbox{$^\circ$}$ at a distance of 100 $\mu$as. The exact location of the component varied with the a priori models and the authors suggest that the map should be taken as a strong indication of a central core with a jet extending out from the core, rather than a full hybrid map of the source. The hybrid maps strongly suggest that the jet twists drastically going from $\mu$as to mas scales.

3.15 CTA102

The arc-second scale structure of the source is dominated by a central core and two other components (Spencer et al. 1989). At $\lambda$18cm the stronger component has a flux density of 0.2 Jy and is located at $\sim$1.6 arcsec in PA $\sim 140\hbox{$^\circ$}$ while the weaker component is only 0.1 Jy at 1.0 arcsec in PA $\sim -40\hbox{$^\circ$}$. Both components have a steep spectrum. Observations at $\lambda$6cm (Wehrle & Cohen 1989), at $\lambda$18cm, and at $\lambda$1.3cm (Rantakyrö et al. 1996) show a central double knot feature (separated in NS direction by $\sim$3mas) with an extended diffuse tail bending sharply to the SW. Rantakyrö et al. (1996), showed that the major contribution to the variability at wavelengths shorter than 32cm is intrinsic to the source.

  

Figure 15: CTA102 100GHz map from 1990 April 22. Peak flux density = 2.5 Jy/Beam. Contour intervals are chosen as (-3.0, 3.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0) $\times$ 10 mJy/beam. The restoring beam was a elliptical Gaussian (0.44 $\times$ 0.03 mas in PA $=4.1\hbox{$^\circ$}$)

Although we have only have one scan with closure phase information it was sufficient to make a rough hybrid map. The 1990 April map at 100GHz (Fig. 15) resolves the northern component of the central double knot into two components. Due to the low declination of the source and the very poor UV-coverage the beam is very strongly elongated in the NS direction. We have chosen to present the image with the original beam since the convolved beam is markedly different from a circular Gaussian of 50 $\mu$as. The side lobe pattern results from the poor UV-coverage of the observations. The result from model fitting Gaussian components to the UV data can be found in Table18.

   Table 18: Model parameters for Gaussian components fitted to the CTA102 UV data. Explanation of the parameters can be found in Table5. Since it is unclear what is the core in this image we give the position of the components relative to the phase center of the map. The large errors in $\delta$ are attributed to the heavily extended beam due to the very poor UV-coverage

The model fit confirms the central structure with two strong components separated in an east west direction. We are unable to determine which of the components is the core, thus we cannot say how this structure is connected with the structure seen with mas resolution. It is clear that the jet twists and turns at it leaves the core. At $\lambda$3mm the components appears to be separated in an EW direction while at $\lambda$1.3cm the jet direction is NS and later the jet is shown to turn sharply to the E (Wehrle & Cohen 1989; Rantakyrö et al. 1996). Having only one epoch of observations with very limited UV-coverage we cannot say anything about the possible proper motion of components in this source; further observations are needed to investigate this.

3.16 3C454.3

This strong radio-source is an optically violent variable with high optical polarization. It is a strong $\gamma$-ray and X-ray source (Hartman et al. 1993). It has been observed extensively at lower resolutions (Pauliny-Toth et al. 1987) and at 43GHz (Kemball et al. 1996). The mas structure is a typical core-jet structure with the extended jet in PA $\sim -65\hbox{$^\circ$}$(Padrielli et al. 1986). At $\lambda$7mm the core is unresolved and there is a weaker component at PA $\sim 100\hbox{$^\circ$}$ at 0.5mas distance. $\lambda$3mm VLBI observations made in 1993 and 1994 (Krichbaum et al. 1996a) show a central core with an extended jet in PA $\sim -70\hbox{$^\circ$}$. Components in the jet are seen to be moving with $\beta_{\rm app}\sim 6-7$c.

We found fringes only for one scan (6min of data), not enough to make a hybrid map or a proper Gaussian model fit to the UV data. The baseline between Hat Creek and Kitt Peak had a flux density of $\sim$7Jy.