Extended Ionized Gas Clouds in the Abell 1367 Cluster^∗

Three fields of Abell 1367 were observed with Suprime-Cam (Miyazaki et al. 2002) at the Subaru Telescope (Iye et al. 2004) in 2014 April–May in UTC. The field position and orientation are shown in Figure 1, and the observations are summarized in Table 1. The pixel scale of the data is 0202 pixel⁻¹. The total survey area is 2207 arcmin², which corresponds to 1.7 Mpc². We dithered to fill the gap between CCDs by 1.1 arcmin, typically by five exposures.

Zoom In Zoom Out Reset image size

Table 1.  Summary of Mosaicked Images and Observation Log

Filter	PSF Size (FWHM)	${\mathrm{SB}}_{\mathrm{lim}}$(ABmag arcsec−2)a	Date(UT)	Exposure
B	07	28.5	2014 May 01	8 × 600 s
R	12	27.9	2014 Apr 30	5 × 60 s
			2014 Apr 30	10 × 300 s
			2014 May 01	5 × 300 s
i	11	27.4	2014 May 01	10 × 180 s
NB	07–09	27.7	2014 Apr 30	15 × 900 s
			2014 May 01	5 × 900 s

Note.

^a ${\mathrm{SB}}_{\mathrm{lim}}$ represents the 1σ fluctuation of the surface brightness measured in random apertures of 2 arcsec diameter.

Download table as:  ASCIITypeset image

We used three broadband filters (B, R, and i), and a narrowband filter (N-A-L671, hereafter NB). The NB filter was originally designed for observing Hα emitting objects in the Coma cluster at z = 0.0225, and has a bell-shaped transmittance with a central wavelength of 6712 Å and a FWHM of 120 Å (Yagi et al. 2007; Yoshida et al. 2008). Since the redshift of Abell 1367 (z = 0.0217) is comparable to the redshift of the Coma cluster, the filter is also capable of observing Hα in Abell 1367.

The data reduction was mostly the same as Yagi et al. (2010): overscan subtraction, flat-fielding, distortion correction, background subtraction, and mosaicking were performed. The sky background was subtracted with a mesh size of 512 pixels (1.7 arcmin) square. We also applied several improved processing steps. Crosstalk was corrected by the method by Yagi (2012), blooming was masked automatically, and some optical ghosts were corrected by new algorithms (Yagi et al. 2015b). For the reference of the bright stars in the field, the PPM-Extended (PPMX) catalog (Röser et al. 2008) was used to obtain the position at the epoch of the observation, after world coordinate system calibration using astrometry.net (Lang et al. 2010). The star positions were used for the ghost correction.

The flux zero-point was calibrated using the Sloan Digital Sky Survey III (SDSS-III) Ninth Data Release (DR9) photometric catalog (Ahn et al. 2012). The color conversion procedure was the same as in Yagi et al. (2013b), and the color conversion coefficients are given in Yoshida et al. (2016). About 300–900 stars were used for the calibration, and the root mean square (rms) was 0.04–0.05 mag.

We measured the flux in 2'' apertures at 10⁶ random positions in the combined images and estimated the rms from the median of the absolute deviation (MAD) as rms =1.4826 × MAD. In the combined images, northeast and southwest regions have no data, and we discarded the apertures in the regions. As a result, valid aperture positions were $\sim 7\times {10}^{5}$. The calculated rms is given in Table 1 as a limiting surface brightness (SB).

We adopted the Galactic extinction values of 0.084, 0.050, 0.050, and 0.039, for B, R, NB, and i bands, respectively, from the NASA/IPAC Extragalactic Database (NED)⁷ , which uses the Schlafly & Finkbeiner (2011) recalibration of Schlegel et al. (1998) data. The same extinction correction was applied to the whole field.

To detect Hα clouds, we subtracted the R-band image from the NB image. Here, we refer to the R-band subtracted NB-band image as NB–R image, or simply Hα image when it is not confusing. Note that this NB–R image includes [N ii] emission and possible residual flux of the continuum. Following Yagi et al. (2010), we adopted $R\mbox{--}{\text{}}{NB}=0.065$ for the R-band subtraction from the NB image. The color determined the relative flux scaling between R and ${\text{}}{NB}$ for a typical continuum spectrum of objects without Hα emission/absorption at z ∼ 0.02.

If the underlying continuum is bluer, the R–NB is smaller (Figure 3 of Yagi et al. 2010). An excess recognized in NB–R image is thus not always an emission, but could also be a residual of a red continuum. The Hα flux estimation is largely affected by the continuum subtraction. In this study, we therefore focused on the excess in the NB–R image where continuum is negligible. Such an excess would be the ionized gas out of galaxies, EIG. Meanwhile, a quantitative discussion about star formation in the galactic disk, for example, is beyond the scope of this paper. Later we investigate possible star formations in EIGs, but we caution that a future spectroscopic confirmation is necessary (e.g., Yagi et al. 2013a).

The seeing sizes were estimated using SExtractor (Bertin & Arnouts 1996) version 2.19.5. The NB image has a seeing size variation among the field: 07 at NW and 09 at SE. The seeing size of most NB data is smaller than that of the R-band images. To detect extended Hα clouds out of galaxies, the difference of the seeing size is little affected. The oversubtraction of R-band is small around pure Hα emissions, while it is large around objects with continuum (e.g., foreground stars).

In the Hα (NB–R) image, extended Hα clouds were searched for by visual inspection. Then B-, R-, and i-band images were used to reject possible residuals of the bright continuum. In Table 2 we list the newly detected Hα emitting clouds extending beyond galaxies. Previously known EIGs (CGCG 097-073, CGCG 097-079, CGCG 097-087, CGCG 097-087N, and BIG) are also shown as a reference.

Table 2.  Extended Hα Clouds in Abell 1367

ID/Parent	Size	Area	$\langle \mathrm{SB}\rangle$	Redshift	Hα Flux	Mass	Type
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
orphan1	33 × 13	128	26.6 ± 0.1	0.0217a	(3.4 ± 0.2) × 10−15 a	(5.4 ± 0.3) × 108 a	3
orphan2e	36 × 5	27	26.7 ± 0.1	0.0217a	(4.0 ± 0.2) × 10−16 a	(3.1 ± 0.2) × 107 a	3
2MASX J11443212
+2006238	>33 × 13c	>177c	23.9 ± 0.3	0.02402	>(6 ± 2) × 10−14 c	>(2.8 ± 0.9) × 109 c	2
CGCG 097-092	35 × 11	91	23.2 ± 0.3	0.02157	(5 ± 2) × 10−14	(1.6 ± 0.5) × 109	3
(without core)	23 × 10	68	26.9 ± 0.1	0.02157	(1.4 ± 0.2) × 10−15	(2.2 ± 0.3) × 108
CGCG 097-093	31 × 19	232	24.9 ± 0.5	0.01633	(4 ± 3) × 10−14	(3 ± 2) × 109	1
CGCG 097-122	34 × 28	380	23.8 ± 0.5	0.01824	(1.5 ± 0.8) × 10−13	(1.0 ± 0.5) × 1010	1
(BIG+tail)	334 × 145	3306	25.0 ± 0.5	0.0275b	(5 ± 3) × 10−13	(8 ± 5) × 1010	⋯
(tail only)	184 × 45	486	26.7 ± 0.5	0.0275b	(2 ± 1) × 10−14	(3 ± 2) × 109	⋯
CGCG 097-073	84 × 28	559	24.3 ± 0.2	0.02429	(1.3 ± 0.3) × 10−13	(1.0 ± 0.2) × 1010	1
CGCG 097-079	135 × 17	285	23.2 ± 0.1	0.02342	(1.7 ± 0.2) × 10−13	(5.2 ± 0.6) × 1010	1
CGCG 097-087	166 × 20	910	23.1 ± 0.2	0.02218	(5.8 ± 1.1) × 10−13	(2.3 ± 0.4) × 1010	1
CGCG 097-087N	12 × 11	72	24.7 ± 0.6	0.02516d	(1.2 ± 0.8) × 10−14	(8 ± 6) × 108	1

Notes. (1) Name of the cloud. (2) Size of the minimum bounding rectangle of an isophote of 2.5 × 10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻². The unit is kpc × kpc. (3) Projected area in the isophote. The unit is kpc². (4) Mean surface brightness in the isophote. The unit is mag arcsec⁻². (5) Redshift used to calculate the Hα flux. Taken from SDSS DR12 (Alam et al. 2015) unless otherwise noted. (6) Hα flux. The unit is erg s⁻¹ cm⁻². (7) Mass of the cloud. The unit is $\sqrt{{f}_{v}}\,{M}_{\odot }$, where f_v is a volume-filling factor. (8) Morphological type (see Section 4.4.).

^aThe parent galaxy is uncertain, and the cluster redshift is used. The derived Hα flux and mass would have large error. ^bFrom Gavazzi et al. (2003b). ^cThe object is near the edge of the observed region, and the full extent of the cloud is uncertain. ^dFrom Iglesias-Páramo et al. (2002). ^eSF blobs are not included.

Download table as:  ASCIITypeset image

In the first inspection, we did not match the point-spread function (PSF) because the NB images have a better resolution than the R-band images and PSF matching would blur the fine Hα structures. A slight debris of background subtraction sometimes mimics an Hα emitting object, but the high spatial resolution of the Hα image enables us to distinguish artifacts from EIGs. Moreover, a blur of R-band images due to the PSF mismatch does not affect the visual inspection since we focused on objects with a weak continuum flux. In our previous study of the Coma cluster (Yagi et al. 2010), all the EIGs detected with the same procedure as in this study are spectroscopically confirmed to be real EIGs in the Coma cluster (M. Yoshida et al. 2017, in preparation).

As the exposure time and the transparency of the atmosphere were not uniform in our data, the signal-to-noise ratio (S/N) should vary among the positions by ∼10%. In this study, however, the detection of extended Hα clouds was not affected by the S/N variation. As a test, we made a low S/N image, half of the original, by adding artificial noise, and searched for EIGs again. We can find the same set of the EIGs in the lower S/N image, although the detail structures of EIGs were buried in the noise in the by-eye inspection. Because we adopted a threshold of a constant SB in the net Hα and focused on extended features, the measurement of the EIGs is not expected to be greatly affected by the change in S/N, either. For another test, we performed a visual inspection in PSF-matched images and confirmed that the detected objects were unchanged. In the following analyses and figures, the PSF-matched images were used unless otherwise noted.

Pixel values (renormalized counts) in NB–R image are proportional to the Hα SB. We calculated a pixel value that corresponds to Hα SB of 10⁻¹⁶ s⁻¹ cm⁻² pixel⁻¹ and converted the NB–R counts into the Hα SB. As the NB filter has a bell-shaped transmittance curve, the response to Hα emission is dependent on the redshift. In order to take the redshift effect into account, we construct a simple model spectral energy distribution (SED) for each cloud assuming its redshift and measure the model magnitude with the NB- and R-band filters. The filter model includes CCD quantum efficiency, transmittance of the optics, and the atmospheric transmittance (Yagi et al. 2013b).

Figure 2 shows the redshift distribution of galaxies taken from the SDSS DR12 spectroscopic catalog (Alam et al. 2015). Galaxies brighter than r = 17.8 mag and within 15 from the cluster center are counted. We adopted R.A.(J2000) =11^h44^m365, decl.(J2000) = +19°45'32'' for the cluster center (Piffaretti et al. 2011). Based on Figure 2, we defined the galaxies with $0.014\lt z\lt 0.030$ as the member galaxies of Abell 1367. From the distribution, the standard deviation of the recession velocity is estimated as 815 km s⁻¹. The maximum change of the transmittance of the NB filter for the Hα line is about 0.6 mag (factor ∼0.6) in $0.014\lt z\lt 0.030$. The result of the Hα SB and flux in the following analyses may suffer ∼40% error if the redshift of the emission is uncertain.

Zoom In Zoom Out Reset image size

Spectra of EIGs in previous studies (Yoshida et al. 2004, 2012; Yagi et al. 2007) show a wide variety in [N ii]/Hα ($-1.1\lt \mathrm{log}($[N ii]/Hα$)\lt 0.0$). The [N ii]/Hα around BIG given by Cortese et al. (2006) also varies within $0.09\,\lt$ [N ii]/Hα $\lt \,0.42$. In this study, we adopted log([N ii]/Hα) =−0.4, i.e., [N ii]/Hα = 0.4, for model SED for the conversion. The systematic error by this [N ii]/Hα uncertainty is ≲60%. Other emission lines in R band, [S ii], and [O i] are added in the R-band model magnitude assuming log([S ii]/Hα) = −0.4 and log([O i]/Hα) = −1.0, according to Yoshida et al. (2012). Possible contamination of [S ii] in the NB band in lower redshift is also taken into account. The transmittance and the model SEDs at different redshifts are shown in Figure 3.

We also estimated the oversubtraction of the Hα flux in the R band using the model SED at each redshift. Because our continuum subtraction is not performed in magnitude but in the renormalized counts, the oversubtracted flux is proportional to the NB–R pixel value regardless of the Hα equivalent width. In the redshift range, the oversubtraction is 18%–30%. It is comparable to the estimation in a z = 0.023 galaxy by Yoshida et al. (2016) (23%). The redshift dependence is shown as Figure 4. As expected from Figure 3, the pixel count in R band is almost constant, while that in NB band changes according to the redshift. The oversubtraction is corrected when converting NB–R counts into the Hα SB.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We estimated the error of the Hα flux of EIGs in two steps. One is the uncertainty in NB–R count with a calibrated zero-point, and the other is the uncertainty in the conversion of the NB–R count to the Hα flux.

The limiting SB is shown in Table 1, which is a 1σ fluctuation of the SB measured in an aperture of 2 arcsec diameter. The value includes the photon noise of the background sky, the readout noise, and errors in the data reduction. The effect of a possible remnant of an optical ghost is also included as an average value. As the adopted isophote in this study was comparable to or larger than the 1-σ fluctuation and the area of the object is much larger than that of a 2 arcsec circle, the S/N of the isophotal flux is high. Even for the faintest object, the estimated relative error of the count is three percent.

The absolute calibration of NB and R bands relies on the SDSS photometry and the color conversion models. The conversion may have ∼0.04 mag error in each band (Yagi et al. 2013b). The residual of the continuum of overlapping objects (e.g., the disk of the parent galaxy) may remain in the NB–R counts. Yagi et al. (2010) reported a change in R–NB magnitude of the continuum of about 0.15 mag peak-to-peak. This means that the error in the NB–R count could be ∼10% of the rescaled R-band count. In the regions where the rescaled R-band count is comparable to or larger than the NB–R counts, this error is the dominant one. We summed the rescaled R-band count where the count is larger than NB–R counts, and used 10% of the sum as the estimated error of the NB–R counts. The errors given in Table 2 include the errors described above; the background variation, the zero-point uncertainty, and the possible continuum error, assuming that the three errors are independent.

In the conversion of the NB–R count to the Hα flux, various uncertainties exist, since we do not have spectroscopic information of EIGs. As shown in Figure 3, the redshift should change the observed count in NB band. Thus the error of the redshift can introduce an error in Hα flux. As mentioned in Section 2.5, the error is ∼40% if no parent galaxy is identified. Even if the redshift of the parent galaxy is known, the EIG may have a recession velocity different by several hundred km s⁻¹ from the parent, as in the Coma EIGs (Yoshida et al. 2012). If the offset is 300 km s⁻¹, for example, it may introduce ≳10% error. In the conversion from NB–R to the Hα flux, we adopted a model SED with several assumptions in emission line ratios; [N ii]/Hα, [S ii]/Hα, and [O i]/Hα. The uncertainty in[N ii]/Hα directly affects the conversion, which may introduce an error of ∼40%. The uncertainty in [S ii]/Hα and [O i]/Hα only affects the correction of the oversubtraction, which would be ∼10% at most.

The "orphan clouds" appear to be floating in intergalactic space (Figure 6).

Zoom In Zoom Out Reset image size

If the clouds are in Abell 1367, the size of the north clump (orphan 1) is 33 × 20 kpc, and the southeast clump (orphan 2) is 12 × 3 kpc. Although the isophote of 2.5 ×10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻² at z = 0.0217 separates the two clouds, a faint filamentary cloud is connecting the clouds and possible SF blobs are found between the two, as seen in the left panel. It is uncertain whether they are physically related to each other.

At least within 80 kpc, there are no possible parent galaxy candidates. The nearest giant (${M}_{r}\lt -17$) galaxy is CGCG 097-102S, whose projected distance is ∼80 kpc (Figure 7). No such parent-less clouds were found in the Coma cluster (Yagi et al. 2010) or in Abell 851 (Yagi et al. 2015a). In Abell 1367, the tip of the EIG of CGCG 097-083 is ∼170 kpc in the projected distance from the center of the parent galaxy. The tip seems connected to the parent galaxy as a long tail, however.

Zoom In Zoom Out Reset image size

Several questions arise about the orphan clouds: what is the ionizing source, and what is the origin of the gas. The ionizing source is a great problem for all EIGs. Possible mechanisms are a star formation in situ, illumination of the young stars in the parent galaxy, illumination of an AGN, inside shock heating, UV from ambient hot gas, etc. Since the orphan clouds are at least 80 kpc away from the giant galaxies, a UV from a parent galaxy disk is unlikely. A possible star formation around the cloud is seen in some of the clouds. In Figure 6, Hα emission and a blue continuum is recognized near orphan 2, which are marked as SF blobs.

The region is also visible in a Galaxy Evolution Explorer (GALEX) near-ultraviolet (NUV) image in the archive.⁸ If the SF blobs are the ionizing source of the whole clouds, however, a gradient of Hα SB is expected. The SB should be bright near the region and fainter in distant regions. No such gradient is seen in the data, and thus ionization by the SF blobs is unlikely. More details of the SF blobs are discussed in Section 5.2.

To examine the possibility of AGN illumination, we compared them with an example of a known AGN-illuminated cloud, Hanny's Voorwerp (Lintott et al. 2009; Keel et al. 2012) at z = 0.050. Hanny's Voorwerp is a famous extended ionizing cloud in a less crowded environment found in SDSS imaging data by the Galaxy Zoo project (Lintott et al. 2008). The ionizing source of Hanny's Voorwerp was revealed to be the AGN of the parent galaxy (IC 2497) 50 kpc away. The size of Hanny's Voorwerp (18 × 33 kpc) is comparable to that of orphan 1 ($\sim 20\times 33\,\mathrm{kpc}$), but Hanny's Voorwerp is much brighter than the orphan clouds. According to Table 2 of Lintott et al. (2009), the mean SB of Hα+[N ii] is $5\times {10}^{-16}$ erg s⁻¹ cm⁻² arcsec⁻², which corresponds to 21.7 mag arcsec⁻² in our NB–R image. The mean SB of the orphan clouds is 26.6 mag arcsec⁻², about 1/90 in flux. The nearest AGN from the orphan clouds is CGCG 097-121 at ∼180 kpc in Gavazzi et al. (2011), which is classified as a LINER. If the orphan clouds are also ionized by the AGN, the ionizing flux from the AGN would be weaker by an order of magnitude or more than that at Hanny's Voorwerp because of the distance. Added to that, if the ionizing flux of CGCG 097-121 is an order of magnitude weaker than the source AGN of Hanny's Voorwerp, the mean SB of ∼27 mag arcsec⁻² of the orphan clouds could be reproduced. Thus the AGN illumination is one promising scenario. Shock heating and UV from the hot gas are also promising ionizing mechanisms.

The origin of the gas is another question. They would have been ejected from a galaxy, but the mechanism is unclear. It can be a ram-pressure stripping, a jet, a wind, a tidal interaction, etc. The clouds are at least at a distance of 80 kpc from their parent if their parent is a giant galaxy. Assuming that their speed to the parent galaxy is ≲1000 km s⁻¹, they should have survived in the cluster environment for ≳70 Myr. As no counterpart is found near the clouds in the H i survey of Abell 1367 (Scott et al. 2010), it is rather unlikely that the orphan clouds are reionized portions of a long tail, such as the long tail of NGC 4388 in the Virgo cluster (Yoshida et al. 2002; Oosterloo & van Gorkom 2005).

Another possibility is that the parent galaxy is a very faint dwarf. A possible ultra-diffuse galaxy (UDG; e.g., Koda et al. 2015; van Dokkum et al. 2015a, 2015b; Yagi et al. 2016) is detected near the possible SF region, and the elongation of the SF blobs seems to be aligned to the galaxy. The galaxy is visible in the bottom left panel of Figure 6, but it shows no Hα emission. At the distance of Abell 1367, the R-band absolute magnitude of the galaxy is ${M}_{R}\sim -14.5\,\mathrm{mag}$ and its effective radius is ∼4 kpc, measured by GALFIT (Peng et al. 2002, 2010).

If the redshift of the UDG is comparable to the SF blobs, the SF blobs may be associated with the galaxy, and it will be an indication that may help to understand the formation of UDGs. Meanwhile, the shape of the whole orphan clouds is unlikely to have been stripped from the possible UDG. There are many other faint dwarfs whose redshift is unknown. Current data are insufficient to investigate the parent galaxy of the clouds in more detail.

CGCG 097-092 (Figure 8) shows an EIG with a cone-like appearance extended to ∼30 kpc. The parent galaxy does not show signs of an AGN (Gavazzi et al. 2011), while the spectrum of the center of the parent galaxy shows a starburst feature. The UV light from the starburst would be the source of the ionization. In Table 2 we also show the property of the tail, without the core region. The origin of the EIG could be ram-pressure stripping or a superwind by the central starburst (Veilleux et al. 2005). We will be able to distinguish the two when the recession velocity profile will become available in the future. The velocity of EIG should be comparable to the parent near the galaxy if the ram-pressure is the mechanism, while a difference should be seen if a wind is the origin.

Zoom In Zoom Out Reset image size

The direction of the tail is somehow "toward" the cluster center; the angle is about 30°. The recession velocity of the parent galaxy shows a small difference from that of the cluster (∼−40 km s⁻¹). If ram-pressure stripping is the origin, this suggests that the parent galaxy moves away from the cluster center in the projected plane.

The EIG of 2MASX J11443212+2006238 was detected at the very edge of our field Figure 5. The EIG may have a much longer extension. In Figure 9, residuals of CCD chip edge pattern remain. The sharp cutoff on the left would be an artifact. Thus the size and total Hα flux are unreliable. Meanwhile, the S/N of the EIG is high enough, and the mean SB is reliable.

Zoom In Zoom Out Reset image size

SDSS spectrum of the parent galaxy shows clear Hα emission from the core, and Gavazzi et al. (2011) classified it as Hii. Kriwattanawong et al. (2011) also reported Hα from the galaxy.

Interestingly, the parent galaxy is red, as shown in Section 4.2. We checked the g − i color in Consolandi et al. (2016) and confirmed that g − i is also red ($g-i=1.03$). In the NB–R image, the central ∼15'' region shows an excess, which may be the remaining star formation, although the NB–R color is unreliable for an Hα estimator in the galactic disk. In the disk, no obvious spiral pattern is recognized in Hα. In the Hα image, the southern half of the galaxy shows a sign of negative NB–R, while it is positive in the northern half. This morphology resembles IC 4040 in the Coma cluster (Yoshida et al. 2008) and implies that 2MASX J11443212+2006238 suffers strong ram-pressure stripping from the south direction. The red color and negative NB–R may be implying that this galaxy is partly in a transition phase from an SF galaxy to a post-SF galaxy.

We found that the Hα tail from BIG continued about twice longer than reported by Cortese et al. (2006) and reaches a projected distance of ∼330 kpc (Figure 10). The tail was called "NW" in Cortese et al. (2006).

Zoom In Zoom Out Reset image size

In Table 2 we list two measurements. We first measured the whole Hα including BIG regions, Then, we set the northwest boundary of BIG as the knot named K1 in Gavazzi et al. (2003b) and Cortese et al. (2006), and measured the tail out of the boundary. The total Hα gas mass of BIG+tail is calculated to be $(8\pm 5)\times {10}^{10}\sqrt{{f}_{v}}\,{M}_{\odot }$, where f_v is the volume-filling factor. Without the Hα around BIG, the tail mass would be $\sim (3\pm 2)\times {10}^{9}\sqrt{{f}_{v}}\,{M}_{\odot }$. The long tail implies that the group would have suffered ram-pressure from the intracluster gas of Abell 1367 for several hundred Myr, while the group has moved roughly straight toward the southeast.

The ionizing source of the long Hα tail is a great mystery in EIGs (e.g., Yoshida et al. 2004; Cortese et al. 2006; Yagi et al. 2007; Kenney et al. 2008). Boselli et al. (2016) speculated that the ionizing mechanism of 80 kpc long tail of NGC 4569 in the Virgo cluster could be shocks, magnetohydrodynamic (MHD) waves, or heat conduction. The discovery of the 330 kpc long Hα tail of BIG sets constraints on possible models. In B, R, and NB composite images, we detected several SF blob candidates. The zoomed-in images are shown in Figure 11. All of them have counterparts in GALEX NUV data. Although they may ionize some gas in the vicinity, they are insufficient to ionize the distant part of the tail. The details of the SF blobs are investigated later in Section 5.2.

Zoom In Zoom Out Reset image size

As seen in Figure 10, the tail of BIG shows a sign of winding with a width of ∼40 kpc. The morphology might be explained as the winding results from the internal motion of the galaxy in the group. Although a merger and interactions among the members of BIG has been suggested (Cortese et al. 2006), here we calculate a simple model. Assuming that the typical velocity of group member galaxies around the center of the group is ∼200 km s⁻¹ (Gavazzi et al. 2003b), the crossing time of 40 kpc would be ∼200 Myr. In the Hα image, one or two possible periods are barely recognized in the tail. The projected length of 330 kpc tail might have been formed in 400–800 Myr, and the tangential velocity of BIG would be 400–800 km s⁻¹. As a result, the Hα tails form helices, and we see them as winding shapes by projection. This speculation can be examined by measuring the velocity profile of the tail along the winding by future spectroscopic observations.

CGCG 097-093 (Figure 12) shows a morphology like jellyfish galaxies (e.g., Cortese et al. 2007; Smith et al. 2010; Ebeling et al. 2014; Poggianti et al. 2016). Enhancement of faint blue stars and Hα emission are seen to the northeast. Inside of the disk, an asymmetry of the spiral arms is recognized, which may be the sign of a recent tidal interaction. The recession velocity of CGCG 097-093 is 7298 km s⁻¹. Its southwest neighbor, CGCG 097-088 is at 5616 km s⁻¹, and the north neighbor CGCG 097-094 is at 7998 km s⁻¹. If the Hα ejection is affected by interaction with the neighbor, it would be with CGCG 097-094.

Zoom In Zoom Out Reset image size

Meanwhile, the morphology in Hα suggests a ram-pressure origin, as the galaxy-wide gas flow would be difficult to be caused by tidal interaction alone. Although the asymmetric stellar distribution suggests a tidal interaction, some ram-pressure stripped galaxies also show an asymmetry (Abramson & Kenney 2014; Jáchym et al. 2014). Vollmer (2003) suggested a mixture of tidal interaction and ram pressure for NGC 4656 in the Virgo cluster, which shows an extended gas tail and an asymmetric stellar distribution.

CGCG 097-122 (NGC 3859) looks like a spiral galaxy seen edge-on with a distorted faint halo that was revealed by our deep imaging (Figures 13). The halo is elongated toward the northeast almost along the major axis of the galaxy. At the southwest side of the galaxy, the halo is extended to a projected distance of ∼10 kpc above the galaxy disk. North of the galaxy, a small galaxy that is possibly an interacting dwarf companion is seen. A part of the extended halo may be formed by the interaction between this companion and CGCG 097-122.

Zoom In Zoom Out Reset image size

The bright disk Hα emission together with the very blue color of the galaxy and the emission line dominated nuclear spectrum indicate that the star formation of this galaxy is very active. Its nuclear spectrum is typical of starburst galaxies (Gavazzi et al. 2013) and has no sign of AGN activity (Gavazzi et al. 2011). The total Hα+[N ii] flux of CGCG 097-122 is ${10}^{-12.65}$ erg s⁻¹ cm⁻² (Gavazzi et al. 2006), and [N ii]/Hα = 0.36 (Gavazzi et al. 2013). The Hα luminosity is calculated to be $\sim 1.9\times {10}^{41}$ erg s⁻¹, which is slightly higher than the typical value of the Hα luminosity of nearby starburst galaxies (∼(5–10) × 10⁴⁰ erg s⁻¹; e.g., Strickland et al. 2004).

The morphology of the Hα emission is peculiar (Figure 13). It is characterized by its two narrow spurs whose widths are ∼2 kpc extended ∼10 kpc northwest and southeast from the galaxy disk, and strong disk Hα emission. The spurs look like two wings of a flying bird. The north part of the Hα nebula of CGCG 097-122 seems to be flowing toward the northeast. In addition, the two spurs also are bending in the same direction. These morphological characteristics suggest that the hot gas of the galaxy is pushed by the ram pressure from the southwest. However, in contrast to the extended nebulae seen around other galaxies in this study, there is no clear indication of one-sided elongation of the ionized gas. There is no sharp cutoff at the southwest edge of the Hα nebula, and no extended part of Hα out of the galaxy at the northeast, but it rather shows an extension toward east-southeast. In addition, the northern faint flowing part of the nebular continuum is overlapped by the dwarf companion mentioned above, suggesting that this part of the nebula could be formed by a galaxy–galaxy interaction rather than ram pressure. Therefore, the ram-pressure stripping may not be strong enough to create the overall structure of the EIG around CGCG 097-122, although it possibly affects the slight bending of the spurs and the positional offset between the bright part of the Hα and the stellar continuum.

The two narrow spurs would be parts of a bipolar bubble driven by the nuclear starburst. It is well known that active star formation drives a large-scale outflow from the SF region; a starburst superwind (Veilleux et al. 2005). In its early phase, the outflow forms a large bubble of ionized gas, a superbubble, then the bubble is broken due to a continuous energy input from the starburst (e.g., Strickland & Stevens 2000). The morphology and size of the spurs resemble broken superbubbles observed in nearby starburst galaxies such as NGC 253 (Strickland et al. 2002), NGC 1482 (Hoopes et al. 1999), and NGC 3628 (Fabbiano et al. 1990). Sometimes the superbubble or the superwind has an antisymmetric morphology in the ionized gas emission; one side of the bubble is much brighter than the other side. For example, the superwind of NGC 3628 is seen as a long plume from the galaxy disk (Fabbiano et al. 1990). The large-scale faint Hα nebula of NGC 253 is characterized by the ∼9 kpc long bidirectional plumes extending from the disk (Strickland et al. 2002). The morphology and size of the two spurs of CGCG 097-122 are quite similar to these features, suggesting that the spurs originate in a starburst superbubble/superwind.

We found that EIGs in Abell 1367 are typically longer than those in the Coma. In Figure 14 we plot the cluster-centric projected distance versus the long side of the bounding box of EIGs in both clusters. This does not mean that the EIG is continuously extended to this length. The length rather approximates the distance of the tip of the EIG from the parent. As the virial radius of the Coma is larger than that of Abell 1367, the difference increases when we normalize it by the cluster virial radius. In Figure 14 the longest one is the tail of BIG. The parents of other long (>50 kpc) EIGs are three bright spirals, CGCG 097-073, 097-079, and 097-087 in Abell 1367, and GMP2559(IC4040), GMP4060(RB199), and GMP2910(D100) in the Coma. All of them are thought to be caused by ram-pressure stripping (Gavazzi et al. 2001a, 2001b; Yagi et al. 2007; Yoshida et al. 2008, 2012). Of these spirals, two in the Coma, GMP4060 and GMP2910, show a post-starburst signature.

Zoom In Zoom Out Reset image size

To investigate the color of the parent galaxies, we first used the SDSS DR12 spectroscopic catalog for the member selection of Abell 1367. Selection criteria were $0.014\lt z\lt 0.030$, $r\lt 17.7\,\mathrm{mag}$, and in the region covered by our observation. We refer to the list as an "unbiased member list." Although the spectroscopic target selection of SDSS is not complete, it is expected to be unbiased. Next, the galaxies in SDSS DR12 photometric catalog whose spectroscopic information is not available in SDSS DR12 were cross-identified with other spectroscopic catalogs (Iglesias-Páramo et al. 2002; Cortese et al. 2003, 2004, 2008; Gavazzi et al. 2003b; Smith et al. 2004; Kriwattanawong et al. 2011), and added if their redshift satisfied $0.014\lt z\lt 0.030$ and $r\lt 17.7\,\mathrm{mag}$. We used our Suprime-Cam data for visual inspection of the cross-identification, and thus only the galaxies in our observed field are added. Although the completeness is higher in the list, the sampling is heterogeneous. We refer to the sample as an "extended member list." The color-magnitude diagram (CMD) of the member galaxies is shown as Figure 15 using the extended member list. We applied the AB magnitude offset adopted by K-correct (Blanton & Roweis 2007) v4, ${m}_{\mathrm{AB}}-{m}_{\mathrm{SDSS}}=0.012(g)$ and 0.010(r), and the Galactic extinction correction (A_g = 0.076 and A_r = 0.053). The color-magnitude relation (CMR) of early-type members of Abell 1367 is

$$\begin{eqnarray}&&g-r=1.210-0.0304r,\end{eqnarray} \tag{ 1 }$$

which is obtained from fitting to the data. When we classify the galaxies into red/blue at 0.2 mag redder than the CMR (broken line in Figure 15), six parents are blue and two are red (2MASX J11443212+2006238 and CGCG 097-122). Regarding the BIG members, CGCG 097-114 is blue, while CGCG 097-120 and CGCG 097-125 are red. SDSS J114501.81+194549.4 is marginally red (by 0.01 mag).

Zoom In Zoom Out Reset image size

The number of members and EIGs are given in Table 5. For the statistical test, we did not count CGCG 097-087N, as it is not included in the unbiased member list. The CMR and color threshold of the Coma is taken from Yagi et al. (2010), and the unbiased member list of the Coma is constructed from SDSS DR12 by applying the criteria that $0.015\lt z\lt 0.035$, $r\lt 17.7\,\mathrm{mag}$, and in the region covered by the observation.

Because the population size is very small, most of the differences are within the statistical error. Only the blue member fraction of Abell 1367 is significantly higher than that of Coma. The whole EIG parent fraction is six percent in the Coma and nine percent in Abell 1367. The difference is marginal (p = 0.06, upper).

Figure 16 shows the distribution of galaxies in the projected distance from the cluster center versus the recession velocity plane. Galaxies with a spectroscopic redshift by SDSS are plotted. Many of the parents have a large difference from the cluster recession velocity. This trend is similar to that in the Coma cluster (Yagi et al. 2010).

Zoom In Zoom Out Reset image size

The two parents near the cluster velocity are CGCG 097-087 and CGCG 097-092. As the tail of CGCG 097-087 points away from the cluster center (Figure 5), it is natural to think that CGCG 097-087 would be entering the cluster directly to the center on the tangential plane. Meanwhile, the tail of 097-092 shows no alignment toward the cluster center (Figure 5).

As in Yagi et al. (2010), the EIGs of Abell 1367 are classified into three types: (1) connected Hα clouds with a disk-wide star formation in the parent, (2) connected Hα clouds without a disk-wide star formation in the parent, and (3) detached Hα clouds. Type 1 includes CGCG 097-073, CGCG 097-079, CGCG 097-087, CGCG 097-087N, CGCG 097-093, CGCG 097-122, and three BIG galaxies (CGCG 097-120, CGCG 097-114, and CGCG 097-125). Type 2 includes 2MASX J11443212+2006238. Although star formation partly remains in the galaxy, it resembles GMP3779 and GMP3896 in Yagi et al. (2010), which were classified as type 2. Type 3 includes CGCG 097-092, SDSS J114501.81+194549.4, and the orphan clouds. The type classification is given in Tables 2 and 3.

In the Coma cluster, the number of each type was 4 (Type 1), 4 (Type 2), and 6 (Type 3). In Abell 1367, the fraction of Type 1 is larger and Types 2 and 3 are smaller. In the Coma cluster, four faint blue parents with detached clouds (Type 3) are found, and they show a post-starburst spectrum. In Abell 1367, there are fewer post-starburst parents. Only SDSS J114501.81+194549.4 in BIG shows signs of post-starburst signature, whose color is barely red (0.01 mag redder than the demarcation line).

Although many statistical properties of EIGs and their parents are similar in Abell 1367 and the Coma, there are several differences. In Abell 1367 the EIGs are longer and tend to be connected to parents. The EIG parents in Abell 1367 have more star formation in the disk, and post-starburst EIGs are fewer. The fraction of EIG parents and blue galaxies is higher in Abell 1367.

Under the assumption that the EIG-parents connection would evolve from Type 1 to Type 3, the larger number of Type 1 in Abell 1367 and fewer Type 3 suggest that the EIGs and parents in Abell 1367 are in an earlier stage of the evolution on average. Most of the EIG parents are thought to be galaxies that infell into the clusters. The gas in the galaxy will be lost due to certain processes, while part of the gas may form an EIG. After some time, the star formation in the galaxy will cease, and the color of the galaxy will eventually become redder. The larger fraction of blue galaxies in Abell 1367 suggests that not only the EIG parents, but also other galaxies are still unprocessed. As the dominant mechanism of the gas removal in cluster environment is thought to be ram-pressure stripping (e.g., Boselli et al. 2016), it is natural that the evolution is slower in Abell 1367. From the X-ray observation by ASCA (Fukazawa et al. 2004), the central gas density of Abell 1367 is about 1/5 of that of the Coma, the mass is about 1/14, and the core radius is 30% larger. Thus the ram pressure, which is proportional to the density and the square of the velocity, should be weaker in Abell 1367. The smaller number of parents with post-starburst signature in Abell 1367 is also explained by the lower efficiency of the ram-pressure stripping.

The longer length of EIGs in Abell 1367 requires another explanation. The length of EIGs shows the distribution of the ejected gas from the parents and/or the reach of the ionization photons. If the ionization mechanism is spatially restricted, such as an UV from the young stars of the parent galaxy or inside the EIG, or AGN, the EIG length would be determined by the range of the flux. If it is the case, however, a gradient of SB according to the distance from the UV source is expected, which is not observed in the EIGs in Abell 1367 and in the Coma. Therefore the length would mainly represent the distribution of the ionized gas. According to GOLDMine (Gavazzi et al. 2003a), the parent galaxies of a long EIG in Abell 1367 have a stellar mass comparable to masses in the Coma. Thus the original gas before stripping would have been comparable as well. We present two possible explanations of the longer distribution of the ionized gas in Abell 1367. One is that the removal of the gas is relatively slower and longer lasting in Abell 1367, because of the low efficiency of the ram-pressure stripping. Another possibility is that the short length of EIGs in the Coma does not mean a short duration of the gas removal, but that the distant part of the gas was heated up and became invisible in Hα (Tonnesen & Bryan 2010; Tonnesen et al. 2011). The length reflects the survival timescale of the ionized gas in the cluster. Although the uncertainty is large, the estimated masses of the ionized gas of EIGs in Abell 1367 are higher than those in the Coma by about an order, while the difference in the length of EIGs is about twice as large. The mean SB in Hα is also brighter in EIGs in Abell 1367. These results suggest that the slow heating scenario would be more likely.

There are several studies on star formation in a stripped tail: inside BIG (Gavazzi et al. 2003b), and galaxies in other clusters (e.g., Cortese et al. 2007; Sun et al. 2007, 2010; Yoshida et al. 2008, 2012; Hester et al. 2010; Yagi et al. 2013a; Kenney et al. 2014). Results from simulations are also reported (e.g., Kapferer et al. 2008, 2009; Kronberger et al. 2008; Tonnesen & Bryan 2010; Tonnesen et al. 2011; Tonnesen & Bryan 2012). Regarding the tail of BIG, Cortese et al. (2006) predicted that star formation may occur since the mean column density is high enough to start star formation.

In the B, R, and NB composite image in this study, we detected possible SF blobs in the tail of BIG and near the orphan clouds (Figures 6, 11, and Table 4.) We checked the GALEX image from the archive, and found that they are also detected in UV. We refer to the candidates as SF blobs for simplicity hereafter, although spectroscopic confirmation is needed to verify that they are genuine SF blobs in Abell 1367. We named the SF blobs in the tail of BIG B1–B6, according to the distance from BIG, and the one near the orphan clouds (orphan 2) we call O1. Their properties are shown in Table 4, and cutouts are shown as Figures 17 and 18. In this section, we mainly used images before PSF matching to make use of higher spatial resolution in NB band. Meanwhile, the magnitude shown in Table 4 was measured in PSF-matched images for consistency. The magnitude and Hα flux were measured within the isophote of 2.5 × 10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻². We assumed z = 0.0275 for B1–B6 and z = 0.0217 for O1. In the B, R, and NB composite of B2, B4, and B5, possible blended fore- and background objects are recognized as yellow objects. Thus their magnitudes are contaminated by the blend. It should also be noted that since the PSF and deblending is different, the magnitude in optical and UV is not directly comparable.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 4.  Candidates of Star-forming Blobs in the Tail of BIG and Near Orphan 2

ID	R.A.(2000)	Decl.(2000)	Hα Fluxa	Bb	Rb	ib	FUV	NUV
B1	11h44m453	+19°49'36''	$3\times {10}^{-16}$	22.8	22.9	23.0	22.43 ± 0.17	22.65 ± 0.26
B2c	11h44m444	+19°49'39''	$5\times {10}^{-16}$	23.0	22.6	22.4	22.90 ± 0.25	22.40 ± 0.24
B3	11h44m422	+19°49'50''	$2\times {10}^{-16}$	23.6	23.9	24.4	21.98 ± 0.18	23.34 ± 0.35
B4c	11h44m394	+19°49'49''	$2\times {10}^{-16}$	23.1	22.5	22.6	22.36 ± 0.21	22.92 ± 0.24
B5c	11h44m403	+19°50'46''	$3\times {10}^{-16}$	23.1	22.6	22.8	22.30 ± 0.19	21.78 ± 0.18
B6	11h44m327	+19°51'24''	$3\times {10}^{-16}$	23.6	23.3	23.7	22.33 ± 0.16	22.67 ± 0.32
O1	11h44m252	+20°09'45''	$6\times {10}^{-16}$	20.5	20.2	19.9	21.79 ± 0.15	21.49 ± 0.11

Notes.

^aHα flux within the isophote of ×10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻². The unit is erg s⁻¹ cm⁻². Assuming z = 0.0275 for B1–B6, and z = 0.0217 for O1. ^bMeasured within Hα isophote, and not directly comparable to UV magnitudes. ^cBlended with red (probably fore- or background) objects.

Download table as:  ASCIITypeset image

The regions are very blue in the UV (FUV–NUV) and in the optical (B–R and R–i); −1.4 < FUV–NUV < 0.6, −0.3 ≤ B–R ≤ 0.6, and −0.5 ≤ R–i ≤ 0.3, before internal extinction correction. The color implies that they are SF blobs and do not suffer heavy dust extinction. The Hα luminosity of SF blobs is estimated to be (2–7) × 10³⁶ erg s⁻¹, although the value has quite a large uncertainty, as discussed in Section 2.6.

5.2.1. SF Blobs in the Tail of BIG

5.2.2. SF Blobs near Orphan 2

The SF blobs near orphan 2 are almost perpendicularly aligned to the Hα filament from orphan 2 (Figure 19). Although the SF blobs are resolved into several compact sources in B and NB images before smoothing, they are merged in PSF-matched images, and we measured the magnitude as a whole in Table 4.

Zoom In Zoom Out Reset image size

In the right panel of Figure 18, several residuals remain in the Hα image. The center panel of Figure 18 shows a B, R, and i composite as a reference; the objects that are orange-yellow in the B, R, and i composite are not SF blobs, and the Hα excess are fake because of a PSF mismatch, and they disappear in the PSF-matched Hα image.

In broadband images, a faint stream is seen near the SF blobs. The B-band image is shown in the bottom left panel of Figure 19. Because Hα is weak along the stream (Figure 18), the stream would be stars and might be a tidal tail. The stream seems to connect two galaxies, SDSS J114425.39+200923.2, and SDSS J114426.00+201001.8 across the orphan 2 filament. Their projected distance is 18.3 kpc if they are at the distance of Abell 1367. At the opposite side of SDSS J114426.00+201001.8, the stream is vague but possibly reaches the UDG, which is about 34 kpc away from the south galaxy (SDSS J114425.39+200923.2). As the three galaxies, SDSS J114425.39+200923.2, SDSS J114426.00+201001.8, and the UDG does not have spectroscopic information, the galaxies may not be a member of Abell 1367. Meanwhile, as the SF blobs show strong excess in Hα, they would be in Abell 1367.

As the SF blobs are aligned to the blue stream perpendicular to the Hα filament of orphan 2, and they are 4–8 kpc away from the Hα filament, they would be an accidental overlap on the orphan clouds. In Table 2 the flux of orphan 2 therefore does not include the SF blobs.

The two are well known for their prominent tails (e.g., Gavazzi & Jaffe 1987; Gavazzi 1989; Boselli et al. 1994; Gavazzi et al. 1995, 1998, 2001a; Scott et al. 2010). The Subaru images as Figure 8 is shown as Figure 20. Boselli & Gavazzi (2014) showed a preliminary result from the same data as this paper.

Zoom In Zoom Out Reset image size

CGCG 097-087 (UGC 6697) is also the best-studied galaxy in Abell 1367 (e.g., Gavazzi & Jaffe 1987; Gavazzi 1989; Boselli et al. 1994; Gavazzi et al. 1995, 1998, 2001b; Scott et al. 2010). Gavazzi et al. (2001b) presented an Hα image and spectra of the galaxy and a part of the tail with a detailed investigation.

In the Subaru Hα data (Figure 21), we found that the tail of CGCG 097-087 extends more than >150 kpc from the core, more than twice as long as previously known in Hα (Gavazzi et al. 1995) and in other wavelengths (Sun & Vikhlinin 2005; Scott et al. 2010).

Zoom In Zoom Out Reset image size

In the Subaru Hα data, CGCG 097-087N shows clear twin tails without a stellar counterpart toward CGCG 097-087 (Figure 21). Zoomed-in images before PSF matching are shown as Figure 22. The tails that overlap the CGCG097-087 disk may be an indication that helps to understand the complex velocity field of the CGCG 097-087 disk (Gavazzi et al. 1984, 2001b).

Zoom In Zoom Out Reset image size

CGCG 097-087 shows a quite red color (g − r = 1.81) in the DR12 catalog. We checked the SDSS database to find that the center of the galaxy suffers heavy dust extinction and SDSS extracted the region. The color therefore does not reflect its stellar population. We therefore used the color from the GOLDMine Database (Gavazzi et al. 2003a) for the galaxy, which measured SDSS images with the method by Consolandi et al. (2016). The adopted color of CGCG 097-087 is $g-r=0.61$.

For a statistical discussion of EIG parents, the assignment of the parent galaxy of the complex Hα clouds around BIG is needed. From previous studies, it is known that CGCG097-125 and CGCG 097-114 show a clear Hα tail and were identified as the parents. CGCG 097-120 (v_r = 5609 km s⁻¹; from SDSS DR12) has been thought to be an accidental overlap, since the measured recession velocity of the Hα clouds around BIG is as high as 8000–8800 km s⁻¹ (Cortese et al. 2006), However, a recent spectroscopic observation by MUSE/VLT revealed that CGCG 097-120 is interacting with surrounding Hα gas and shows a smoothly connected distribution of the Hα velocity (G. Consolandi et al. 2017, in preparation). At least these three are therefore parent galaxies of EIGs.

Another possible parent of EIGs around BIG is SDSS J114501.81+194549.4, which shows a clear post-starburst feature from the SDSS DR12 spectrum at v_r = 8220 km s⁻¹. The EIG just south of the galaxy may be connected with the galaxy. The galaxy is thus counted as an EIG parent in Table 5.

Table 5.  Coma and A1367 Statistics in the Surveyed Region

Name	Survey Area (Mpc2)	Member	Blue Member	EIG Parenta	Blue Parenta
Coma	1.2	202	15	12	8
A1367	1.7	120	19	11	6

Note.

^aOnly $r\lt 17.7$ mag ones are counted.

Download table as:  ASCIITypeset image

Yet another possible parent of EIG around BIG is SDSS J114513.76+194522.1 (v_r = 8256 km s⁻¹), about 150 kpc east of the BIG complex. The galaxy shows a sign of stripping toward the west, although no extended Hα feature is seen around it. Moreover, the south clouds of BIG (Cortese et al. 2006) show a sign that they come from far east of BIG. However, this is currently weak evidence. Thus we did not count the galaxy in Table 5 and did not plot it in Figures 15 and 16, but showed it in Table 3 with parentheses as a reference.

Other spectroscopically confirmed dwarfs and knots by Sakai et al. (2002) and Cortese et al. (2006) are fainter than the magnitude limit of the catalog ($r\lt 17.7$) used in the statistical analysis of this study. Previous studies (Sakai et al. 2002; Gavazzi et al. 2003b; Cortese et al. 2006) discussed the possibility that the dwarfs could have been formed in the stripped gas from giant galaxies. The dwarfs would not be parent galaxies of EIGs, but rather children. We therefore do not count them as parent galaxies of EIGs.