PDF《**FULL TITLE**》

239993 such identi?cations. We have also extracted the photometric magnitudes from SDSS (ugriz), the integrated ?ux density measured by FIRST (Sint), the position angle (PA, α) computed from the isophotal distribution in the r-band by SDSS along with the equivalent from FIRST, the major (a) and minor (b) axes measured by SDSS and FIRST, and R50 and R90 measured from the Petrosian intensity pro?le model ?tted to the SDSS images (Petrosian (1976)). We also use the concen- tration c = R90/R50 to help distinguish elliptical galaxies from the others;

for small c the light distribution has an exponential pro?le and ?ts the pro?le of a disk galaxy, whereas for large values of c the pro?le is approximated by a de Vaucouleurs pro?le which ?ts the light distribution of an elliptical galaxy. For our investigation we require galaxies and radio sources for which there are reliable measures of the PAs of extended emission. Since it is di?cult to compute the PA accurately for very faint galaxies and very round ones, we have excluded all galaxies with r >

18, b/a >

0.8 in either SDSS or FIRST, and those with a <

2′′ in FIRST (since FIRST has a 5′′ beam). This leaves a total of

14302 galaxies. We further split the sample based on photometric determined parameters. It has been shown by Strateva et al. (2001) that the colour de?ned by u ? r can discriminate between di?erent distributions in the bimodal g ? r vs. u ? g colour-colour diagram for an earlier SDSS data release. The colour and concentration are known to be correlated (see for example, Strateva et al. (2001)). There is a bimodality in the density of points in the colour-colour diagram for our sample, suggesting that there are indeed two populations, with the redder, more highly concentrated objects being the elliptical galaxies. In order to investigate this quantitatively we have performed a principal component analysis (PCA) on the sample, initially with two variables u ? r and c. The two components which are generated by this procedure are C1 = 0.965c ? 0.262(u ? r) , C2 = 0.262c + 0.965(u ? r) . (1) Dividing at C2 = 3.5 gives a near optimum separation of the two types of galaxy. Jet orientations

3 0

20 40

60 80

1000 1500

2000 0

20 40

60 80

1000 1100

1200 1300

0 20

40 60

80 200

250 300

0 20

40 60

80 800

900 1000 Figure 1. Splits of the data on the basis of PCA and axial ratio. The distributions are: C2 <

3.5 (bottom left), C2 >

3.5 (bottom right), C2 >

3.5 and b/a <

0.6 (top left) and C2 >

3.5 and b/a >

0.6 (top right)

0 20

40 60

80 500

600 700

0 20

40 60

80 260

280 300

320 340

0 20

40 60

80 140

160 180

200 220

240 0

20 40

60 80

40 60

80 Figure 2. Histograms for the case C2 >

3.5 with various splits based on b/a and t ? r. The two plots at the bottom have b/a >

0.6 and the two the top have b/a <

0.6. The two on the left are radio-quiet with t ? r >

?2.5 and the two on the right are radio-loud with t ? r <

?2.5.

4 Browne and Battye 3. Results In Fig.

1 we show the distribution of ?α split by principal component C2 and by axial ratio b/a. As expected the two types of galaxy behave di?erently. It is very evident that the blue, less concentrated galaxies (C2 <

3.5), which broadly represent the disk-dominated, spiral population, have the optical and radio major axes correlated, whereas the red, highly concentrated galaxies (C2 >

3.5), which are part of the elliptical population, have a correlation between the optical major and radio minor axes. The most statistically signi?cant minor axis alignment is obtained for a sub-sample de?ned by C2 >