2.1.Observations, Reduction and Calibration of the Data

The H_α images were acquired at the 1.5m telescope of the Observatorio Astronómico Nacional at San Pedro Mártir (OAN-SPM) in five different observational campaigns from 2002 to 2004. Five narrow band filters were used: three in the hydrogen H_α recombination line, centered at 6570 Å, 6607 Å and 6690 Å, and two for continuum subtraction, with central wavelengths at 6459 Å and 6570 Å. The maximum transmittance of the filter was 70% and they were between 89 Å and 127 Å wide. The integration times were in the range of 1200 and 5400s for hydrogen H_α and 900 and 3600s for continuum, respectively. The air mass was smaller than 1.4 for all but three of the galaxies, and the seeing was between 1.1 and 2.6, with an average value of 1.4. The convolution of the peak transmission of the filters (70%) and the high detector quantum efficiency (90%) allowed us to achieve relatively deep flux limits (7 x 10^-18ergcm^-2s^-1).

The reduction procedure (bias, flat-fields, removal of cosmic rays and sky subtraction) as well as the flux calibration were performed using the ESO-MIDAS software. Three standard stars were observed each night with different air mass in order to perform the flux calibration.

The ultraviolet fluxes were obtained from the 6th release (GR6) of the GALEX database for a total of 30 out of the 36 galaxies of the sample. We used those data with an exposure time of 1500s. Both the FUV and NUV fluxes were determined for the 30 galaxies from the integrated flux provided by the conversion from AB magnitudes: F_v[erg s^-1cm^-2Hz^-1] ECUACION. Full details of the telescope, instruments, calibration and processing pipeline are provided by ^{Martin et al. (2005)} and ^{Morrissey et al. (2005}, ²⁰⁰⁷⁾.

2.2.HII Regions Detection

In order to obtain the SFR, fluxes from the H_α line are needed as well as from the FUV continuum. The simplest approach is to consider the galaxy emission as a whole, e.g. the continuum-subtracted H_α emission. However, as we were interested in a further analysis of the data (such as the luminosity function and the location of the HII regions inside the galaxies) we decided not to use that technique. Instead, we selected the HII regions from the continuum-subtracted H_α images. Only those regions identified by two of us were included in the sample. A circular aperture centered on each region was used to determine the flux. Two different radii were considered for each Hii region. The first one included all the emission of the region and the second one included only the emission with flux values larger than a limiting value, in order to distinguish between proper Hii regions and DIG. As will be discussed in § 6, this limiting flux has been determined by spectroscopic studies and is of 10^-17erg s^-1cm^-2. This procedure was performed manually; the radius aperture just depends on the galaxy distance, which defines the size of the region in the image.

The main disadvantage of this procedure is the difficulty to detect very low luminosity regions. However, as they are very faint, the contribution of their fluxes to the SFR is not significant and will only be important in the study of the luminosity function of the Hii regions inside the galaxies.

In the case of galaxies with large inclinations, it is very difficult to distinguish different regions if they are in the same line-of-sight. This is the situation for 7 of the galaxies in our sample, which have inclinations larger than 70. It can be argued that the inclusion of such galaxies in our sample is not correct because of the difficulty of an accurate extinction correction. However, most of the classical studies of the SFR in galaxies have included high inclination objects without any further correction (e.g. ^{James et al. 2008}; ^{Lee et al. 2009}; ^{Hunter et al. 2010}; ^{Boselli et al. 2015}). In 2018, Wang et al. studied the influence of the inclination on the SFR value and they concluded that the inclination is important only for galaxies more massive than 1010M⊙. Moreover, ^{Buat et al. (2009}) considered as high inclination galaxies only those with inclinations larger than 80. Therefore, we included them because all the galaxies in our sample are less massive than 1010M⊙ and have inclinations smaller than 80; however their SFR values should be considered as upper limits.

A similar problem might appear for those Hii regions which are very close and cannot be separated due to the resolution of our observations.

With the knowledge of the positions of the Hii regions from the H_α images, the GALEX images were inspected to locate the star forming regions and the fluxes from these locations were calculated.

2.3.Flux Corrections

Before the SFR could be determined, the UV and H_α fluxes need to be corrected for extinction. There are contributions to the extinction (i) from the gas inside the galaxy (hereafter, the internal extinction), (ii) due to the intergalactic medium, (iii) due to gas inside our own Galaxy, and (iv) due to the atmosphere. The last one was corrected for the standard way. To correct for Galactic extinction we followed ^{Cardelli et al. (1989)}, using values of R_V from 2.6 to 5.5 (^{Clayton & Cardelli, 1988}) and considering the color excess values from NED.

The most difficult extinction correction is the extinction internal to the galaxy. Since there is no simple way to estimate the extinction coefficient in H_α, A(H_α), this correction is not generally performed. In this article, the determination of the internal extinction coefficient for H_α was computed following ^{Calzetti et al. (2000)}, where they derived an extinction law K(λ) directly from the data in the UV and optical wavelength ranges, as in Calzetti (1997b). As K(λ) is related to the internal nebular extinction, the extinction coefficient in H_α can be obtained as:

A(Hα)=K(λ)×E(B-V).

On the other hand, we followed the formalism of ^{Salim et al (2007)} to estimate the extinction coefficient in FUV, AFUV, as

AFUV=3.32(mFUV0-mNUV0)+0.22,               if (mFUV0-mNUV0)<0.95,3.32,                                              if (mFUV0-mNUV0)>0.95,

where mFUV0-mNUV0 is the UV color of the galaxy in the rest-frame system. The corrected fluxes obtained were of the order of 10^-26erg cm^-2s^-1 with typical values of A_FUV between 0.1 to 2.5 mag.

The internal extinction increased the final fluxes in the galaxies of our sample 15% on average, but the increease was up to 50% for three of the galaxies (UGC 560, UGC 5242, UGC 12212).

Because of the large bandwidth of the filters used in this work (larger than 70 Å), the H_α flux is contaminated with the nitrogen line emission. From our own spectroscopy data of several of the galaxies presented here, the [NIII]/H_a ratio is typically smaller than 0.1 (^{Hidalgo-Gámez et al. 2012}). The flux obtained from our images was then corrected by 10% in order to eliminate this contribution.

3.1.Star Formation Rates of Late-Type Galaxies from H_α Luminosities

From a closer inspection of Column 6 in Table 2, it can be concluded that there are three ranges in the SFR (H_α) values: galaxies with values larger than 1M_ʘyr^-1 (3 galaxies), those with SFR(H_α) between 0.1 and 1M_ʘyr^-1 (9 galaxies), and galaxies with values smaller than 0.1M_ʘyr^-1 (24 galaxies). For the galaxy UGC 9902 the SFR is taken as zero because we could not detect any H_α flux in it. The distribution of the SFR(H_α) values for all the galaxies in the sample is shown in Figure 1 (left). The distribution is not symmetrical, showing a clear asymmetry towards the low SFR values.

Fig. 1 Star formation rate distributions for all the galaxies in the sample from Hα fluxes with DIG (left) and from FUV continuum (right). The range comprised for the Hα fluxes is wider than for the FUV, with the bulk of galaxies having SFR below 0.1 M_ʘyr^-1 for the former and between 1 and 0.1 for the latter 

The galaxy with the largest SFR(H_α) value is UGC 560, one of the most distant in our sample, with a SFR similar to that of the Milky Way (MW) (1.7 M_ʘyr^-1, ^{Robitaille & Whitney, 2010}) while the one with the smallest SFR is UGC 9018, a dS galaxy which is the nearest, one of the smallest in size, and with very few gas left.

As can be seen from Figure 2 (left plots), where the distribution for Sm galaxies is shown in the top row while the one for dS galaxies in the bottom row, the former have larger SFR values than the dS galaxies: a factor of 4.7, according to the average values shown in Table 3 (second column). Actually, only one out of 14 dS galaxies has a SFR larger than 0.1 M_ʘyr^-1. One explanation might be the small number of dS galaxies in our sample. However, they are almost a 40% of the total sample and, therefore, such differences might be due to another reason. We will further discuss this in § 3.6.

Fig. 2 Star formation rate distribution for the Sm galaxies (upper row) and dS galaxies (bottom row). The left plots show the values from the H _α line (SFR(H _α ) _DIG ); the values from the FUV are plotted on the right. 

3.2. Star Formation Rates of Late-Type Galaxies from the UV Continuum

The most interesting feature of the SFR(FUV) values, listed in Column eighth of Table 2 and shown in Figure 1 (right), is the lack of values larger than 1 M_ʘyr^-1. The galaxy with the largest value is UGC 5236, an almost face-on Sm galaxy, with SFR(FUV=0.76 M_ʘyr^-1. Interestingly, the galaxy with the lowest SFR, UGC 8253, is very similar to UGC 5236 in size, inclination and distance. The SFR distribution is not symmetrical, with a deficit of galaxies with low SFRs.

Twenty one out of 30 galaxies have SFR(FUV) values between 1 and 0.1M_ʘyr^-1, (14 Sm and 9 dS) and only 9 galaxies have values lower than 0.1 M_ʘyr^-1. There are differences of 0.11 between the SFR(FUV) of Sm and dS. Although the distribution has a similar range, as can be seen in Figure 2, Sm galaxies have a peak at 0.1 M_ʘyr^-1 while dS galaxies have a smooth distribution.

Star formation rate distribution for the Sm galaxies (upper row) and dS galaxies (bottom row). The left plots show the values from the H_α line (SFR(H_α)_DIG); the values from the FUV are plotted on the right.

3.3.Comparison Between SFR(H _α ) and SFR(UV)

It is well known that SFR(FUV) is larger than the values obtained using the H_α luminosity (^{Bell & Kennicutt, 2001}; ^{Lee et al. 2009}). The UV emission comes from O and B stars with masses larger than 5 M_ʘ, while only stars more massive than 20 M_ʘ can provide the 𝐻 𝛼 emission (^{Werk et al. 2010}). This can be checked in Figure 3, where the SFR(FUV)/SFR(H_α) ratio vs. the SFR H_α values are plotted. Only five galaxies, all of them Sm, are in the negative part of the diagram. The lower the SFR(H_α), the higher the differences between both values. In the earliest stages of the starburst, the emission in H_α is higher and both fluxes are similar. At later stages of the star formation events, the differences between the fluxes in FUV and H_α become larger as the UV continuum decreases more slowly than H_α. The main reason for this is that the UV photons originate from a wider range of stars with longer lifetimes on the main sequence (e.g. ^{Sullivan et al. 2004}; ^{Iglesias-Páramo et al. 2004}).

Fig. 3 Ratio of SFR(FUV)/SFR(H _α ) vs. SFR(H _α ). The solid line indicates when both values are the same. Most of the galaxies show a positive value for this ratio indicating that the SFR obtained from the UV continuum is larger than the SFR from H _α with DIG. 

The values of Table 3 show that the average star formation rates from Hα and FUV are very similar for the sample as a whole, as well as for the Sm galaxies, and there are only large differences for dS galaxies (a factor of 2.7). One explanation might be that the star formation event is older in the latter galaxies; therefore, all the massive stars which can ionize hydrogen have disappeared already, and the Hα flux is quite low. However, there are still a lot of less massive stars which can emit at FUV wavelengths, hence the large value of the SFR(FUV).

3.4.Bars and Star Formation Rates

It has been suggested that strong barred galaxies have larger SFRs (e.g. ^{Sérsic & Pastoriza 1967}; ^{Ho et al. 1997}; ^{Tsai et al. 2013}). However, other autors proposed that bars may decrease the SFR of a galaxy (e.g. ^{Tubbs 1982}; ^{Kim et al. 2018}). ^{Kim et al. (2017)}, using a large sample of galaxies from SDSS, found out that barred galaxies have significantly lower star formation activity than their unbarred counterparts. ^{Ryder & Dopita (1994}) stated that the effect of the bar in the star formation rate is not important. Therefore, the problem is far from being settled yet.

We analized the effect of the existence of bars in the SFR of our sample. Barred galaxies represent 28% of the galaxies in our sample; therefore, the results are significant. Figure 4 shows the distribution of the SFR for barred and non-barred galaxies, with the average values presented in Table 3. Non-barred galaxies have larger SFR than barred ones from 𝐻 𝛼 values, while the SFR is similar when determined with FUV fluxes. However, the distributions are very different in both cases. Barred galaxies have a small range of SFR, with no large values (neither from H_α nor from FUV). The distribution of barred galaxies with FUV values is flat. From our sample it can be concluded that barred galaxies have a lower SFR than non-barred ones.

Fig.4. Star formation rate distribution for barred (bottom row) and non-barred galaxies (top row) in the sample. The left plots show the SFR determined with the H_α luminosity while the right plots show the distribution of the SFR determined with the FUV continuum. Non-barred galaxies have, on average, larger Star Formation Rates. 

How can this influence the lower values of SFR for dS? Actually, the number of barred dS galaxies is small, only three out of 14; therefore, the results cannot be conclusive. However, there are no differences in SFR(H_α) while the differences increase for SFR(FUV). So, the small SFR for dS galaxies must be due to another reason.

3.5.Comparison with Previous Results

There are several studies of SFR involving late-type galaxies that use different diagnostic methods (e.g. ^{Hunter & Elmegreen 2004}; ^{Hunter et al. 2010}; ^{Rosenberg et al. 2008}; ^{Lee et al. 2009}; ^{Buat et al. 2009}; ^{Boselli et al. 2015}; ^{Boselli et al. 2009}; ^{James et al. 2008}; ^{Almoznino & Brosch, 1996}). In general, Sm galaxies have higher values than Im galaxies in all these investigations, with the values for the Im being closer to the SFR of our dS galaxies. On average, the SFR values reported for Sm galaxies are very similar to those we determined when extinction is considered. This is particularly important for the SFR(FUV), because the extinction is larger than in the optical range. This might be the main reason for the differences between our SFR(FUV) values and those reported by ^{Lee et al. (2009)} and ^{Hunter & Elmegreen (2004)}, with the latter authors using the color excess and the extincion law of ^{Cardelli et al. (1989)} and the former ones using the Balmer decrement to determine the internal extinction. Therefore, our values are between two to five times larger than theirs.

Concerning the comparison between the SFR(H_α) and SFR(FUV), our results are similar to other investigations, the SFR(FUV) being higher than the SFR(H_α), except for the KISS sample by ^{Rosenberg et al. (2008)}, where at least 8 out of 19 galaxies have lower SFR(FUV).

Eleven of the galaxies studied here have SFR values previously determined from H_α emission (^{Hunter & Elmegree, 2004}; ^{James et al. 2004}; ^{Lee et al. 2009}; ^{Van Zee, 2001}), and four have previous FUV emission determinations (^{Hunter & Elmegree, 2010}; ^{Lee et al. 2009}). Six of the values for SFR(H_α) are identical to our values, while another five are similar when the uncertainties are considered. The most discordant SFR(H_α) values are for UGC 9018, UGC 8385, DDO 18 and James et al.’s value for UGC 11820. There are several reasons for such differences, among others: the use of different distances or differences in the internal extinction. For example, ^{James et al. (2004)} adopted a constant value of 1.1 mag for the extinction of all their galaxies, regardless of their morphological type or inclinations. This value is higher than the extinction we reported in our § 3. Also, different number of HII regions detected might change the value of the SFR, as in UGC 11820, where in a previous work up to 40 HII regions were considered, and a SFR of 2.9 10-2M⊙ yr^-1 (^{Reyes-Pérez, 2009}) was reported.

3.6.Star Formation Efficiency

Galaxies with distinct spiral arms, like Sc type, have larger SFR due to the fact that they have a large amount of gas and a prominent density wave (^{Kennicutt, 1998}). Late-type galaxies (Sm, Im and dS) have, in general, smaller values of the SFR despite their large amount of gas. However, when the SFR surface density is considered, galaxies behave very differently. ^{Hunter & Elmegreen (2004)}, using a sample of Sm, Im and blue compact dwarf galaxies (BCD), found out that the latter were the galaxies with the largest SFR surface densities (see their Figure 5). A similar result is found here, using the optical radius instead of the disk scale length in V which, according to ^{Hunter & Elmegreen (2004)}, makes no difference for SFR surface densities lower than 0.05 M_ʘ yr^-1kpc^-2. The SFR surface density for dS galaxies is larger (with an average value of 1.6×10-4 M⊙ yr^-1kpc^-2) than that of Sm galaxies (on average 7.7×10-5 M⊙ yr^-1kpc^-2). So, dS are forming a larger number of stars per area than Sm, despite the fact that the latter on average have twice more gas mass than dS (12.5 M⊙ yr^-1 in dS vs. 8.5 M⊙ yr^-1 in Sm).

Fig.5. Correlation between the SFR surface density and the average surface densities of H_I. The dashed lines correspond to constant global star formation efficiencies (1, 10, 100%) from ^{Kennicutt (1998)}. The Sm and dS galaxies are shown with different symbols (stars and diamonds, respectively). The graph at the top shows the KS law when the star formation was determined from the H_α luminosity, while the one at the bottom shows the star formation determined with the FUV continuum. 

In order to study this effect, the well-known Kennicutt-Schmidt law is applied to our data and is presented in Figure 5 for both H_α (top) and FUV (bottom). As no information is available on the amount of molecular gas in these galaxies, the gas density was determined as 1.3×Σ(HI), the ratio between the Hi and the molecular gas being constant (^{Leroy et al. 2005}; ^{Saintonge et al. 2011}). We plot the efficiency lines (defined as the ratio between the SFR and the gas mass) at 1, 10 and 100%, as in ^{Kennicutt (1998)}. From the results of Figure 5, it cannot be concluded that dS are more efficient forming stars than Sm, or vice versa. Most of the galaxies in the sample are near the 10% efficiency line for both calibrators, although FUV values are located mostly between the 10% and 100% lines, while there are H_α galaxies between the 1% and 10% lines. The galaxies UGC 3086 and UGC 3778 seem to be very efficient, close to the 100% efficiency line.

If the figure were divided into three different regions according to the efficiency, the number of Sm and dS galaxies would be very similar in all regions, but at intermediate/lower efficiency the Sm type would be slightly more numerous.

The larger amount of gas in Sm with a similar HI mass might indicate that they will form stars for a longer time than dS, or that they have started to form stars more recently than dS, but we do not have yet enough information to discriminate between these options.

4.1.Concentration Parameter

One parameter that can give us important information about the distribution of the star formation within galaxies is the concentration index (hereafter CI _n ), which is defined as the ratio between the number of HII regions in the inner part and outer part of the galaxy (^{Roye & Hunter 2000}). With this definition of the CI _n , a value of 0 means that all HII regions are in the outer part of the galaxy, and the larger the CI _n the more concentrated towards the center the regions are. In the determination of the CI _n for the 17 galaxies selected before, we follow ^{Roye & Hunter (2000)}; the values are listed in Column 2 of Table 4. All the galaxies have CI _n s larger than 1, and four of them have CI _n values larger than 10, indicating a large concentration of the regions towards the center. Only six of the galaxies have CI values smaller than 4, so their HII regions are located mostly outside the inner part. In general, it can be said that the galaxies in this subsample do not have their HII regions in their outskirts.

Another CI (hereafter, CI _f ) can be defined using the flux of the HII regions (F_in/F_out) instead of the number of them. This is presented in Column 3 of Table 4. Again, large values of CI _f indicate a large amount of flux towards the center, and small values indicate a large amount of flux from the outer regions. Eight of the galaxies have CI _f smaller than 4; therefore their flux comes mostly from the outskirts. On the contrary, five galaxies have a strong luminosity from their center.

The most interesting result emerges when we compare both CIs. One might think that their values should be similar, e.g. if a galaxy has most of its HII regions in the inner part, then most of the flux of the galaxy (in H_a) should come from it as well. Thereupon, it could be said that the galaxy is “well balanced": there is more flux because there are more Hii regions. Such comparison can be done using Columns 2 and 3 of Table 4. Half of the sample has similar values of both concentration index. However, there are three galaxies (UGC 4121, UGC 6151, and UGC 6840) with much higher values of their CI _n (a factor 1.5 or larger) than the CI _f . Therefore these galaxies have a large number of regions inside their central region, but they are not very luminous. On the contrary, four of the galaxies in the sample (24 %) have a CI _f much larger than the CI _n , indicating that, despite being fewer in number, the regions in the center of the galaxy are brighter than those in the outskirts.

It can be concluded that the galaxies in this subsample have a larger number of HII regions in the inner half of the galaxy, and they are also more luminous than the regions in the outer half of the galaxy. However, this is intriguing because in a visual inspection of the galaxies (see Figure 6), the large number of the central HII regions and their dominance of the luminosity is not clear. Therefore, we divided the galaxy into three concentric regions: the inner region with a radius of 1/3 of the total optical radius, the middle region, a torus between 1/3 and 2/3 of the r₂₅, and the outer region, from there up to the optical radius. They are marked with circles in Figure 6. The number of HII regions inside each of these regions, as well as their total flux, are listed in Table 4, Columns 4 to 15. There is no clear pattern, as can be seen in Figures 7 and 8, where the profiles of the number of HII regions (solid line) and the flux (dotted line), both normalized to the total number of regions (or flux) for each galaxy, are shown. The most peculiar profile might be the one of UGC 6840, with fewer regions in the inner annulus, dropping toward the middle section and increasing dramatically at the outer part of the galaxy. Moreover, this galaxy, along with UGC 3989, has the largest differences between the number and the flux distribution. Regarding the non barred galaxies, four out of 12 have the number and the flux of their HII regions concentrated in one of the annuli of the galaxy, while for other four the distribution is very uniform along the galactocentric distance. Four of the five barred galaxies have a large number of HII regions in the middle part of the galaxy. These results do not support the idea that bars inhibit the star formation in the central part (e.g. ^{James & Percival, 2018}; ^{Tubbs 1982}; ^{Kim et al. 2018}) because only DDO 36 has fewer regions inside its central annulus. It is interesting to note that all galaxies with a clear spiral structure have all -or almost all of their star formation regions concentrated in the middle and outer sectors of the galaxy, following the spiral arms. On the other hand, the regions of star formation in galaxies without, or with a not-so-well defined spiral structure, do not show any special distribution.

Fig. 7 Distribution of the number of HII regions (solid line) and their fluxes (dotted lines) along the distance to the center of the regions, for the non-barred galaxies. The y axis is the normalized number of HII regions (solid) or flux (dotted) of the galaxy and the x axis is the location of the galaxy in the any of the three regions we divided the galaxies into: inner, middle and outer part. Each galaxy has a different trend: some galaxies have more regions in the inner part, some others in the outer and some others in the middle. UGC 3989 has its star-forming regions quite homogeneously distributed along the radius, but the flux increases towards the edge. For most of the galaxies, the number of regions and the associated fluxes mimic each other. 

Fig. 8 Distribution of number of H II regions (solid line) and their fluxes (dotted line) along the radius for barred galaxies. Only two of them have most of their regions and flux in the middle part of the galaxy, where the bar is expected to end, and to have an important influence on the star formation. Axis as in Figure 7. From left to right the galaxies are: DDO 36, UGC 8385, UGC 4871, UGC 891 and UGC 6840. 

From these figures and the values in Table 4 it can be concluded that the HII regions of the galaxies in our sample show a variety of distributions, more similar to a random behavior than to a clear pattern, as observed in spiral galaxies.

4.2.Asymmetry of the HII Region Distribution

In a recent study of the dwarf, interacting spiral galaxy IC 1727, it was found that the largest and brightest of its HII regions are facing the companion galaxy, NGC 672 (^{Ramirez-Ballinas & Hidalgo-Gámez 2014}). We aim to check if the HII regions are located in a particular zone in the galaxy, because such an accumulation of the star formation in one particular direction might indicate an interaction. There are two possible approaches: the asymmetry index AI, (^{Roye & Hunter, 2000}), or a simpler one: to just check if there is any preferred locus of star formation inside each galaxy, dividing these into the four orientations, as was done for IC 1727. We will first consider the latter approach. To proceed, the 17 galaxies studied in the previous subsection were divided into four directions: NE, NW, SE and SW (North is up and East to the left). The values of the number of HII regions and the fluxes for each quadrant are listed in Table 5. Half of the galaxies in the sample have a quadrant (or two) with a meaningful increase in the number of HII regions (larger than 35%). This might indicate a previous interaction. Moreover, two of them, both barred galaxies, have the majority of regions in opposite quadrants, while another four galaxies have them in adjacent quadrants. A likely explanation might be that star formation is due to gas coming out along the bar, as seen in other galaxies (^{Sérsic & Pastoriza 1967}; ^{Ho et al. 1997}).

The asymmetry index, based on the number of HII regions (AI_n), is defined as the ratio between the number of regions in the poorest side and the richest side. An AI by number (or by flux, see below) equal to one means that the galaxy is perfectly symmetric along that axis (major or minor axis), that is, we have the same number of HII regions on both sides. On the contrary, an AI of 0 means that all HII regions are concentrated on only one side of the galaxy. We used our images in V and R to determine the major and minor axes and counted the number of HII regions on each side of the axes. We decided to proceed as in ^{Roye & Hunter (2000}), where they a posteriori chose the axis (major or minor) with the largest symmetry. The AI_n values are shown in Column 18 of Table 5. We found that 80% of the galaxies in the sample have an AIn=1 and three of them are perfectly symmetric (AIn=1).

As previously done with CI, an asymmetry index based on the flux can be determined as the ratio between the flux of all the regions on one side and the flux on the other side (AI _f ) (AI _f ). This index is shown in Column 19 of Table 5. Again, our galaxies appear to be very symmetric: only two of them have . According to the AI, most of the galaxies in our sample are predominantly symmetric in both the distribution of the HII regions, and their luminosity.