In this paper , we measure the ellipticities of 30 LSB dI galaxies and compare the ellipticity distribution with that of 80 dEs ( Ryden & Terndrup 1994 ; Ryden et al .
1998 ) and 62 BCDs ( Sung et al .
1998 ) .
We find that the ellipticity distribution of LSB dIs is very similar to that of BCDs , and marginally different from that of dEs .
We then determine the distribution of intrinsic shapes of dI galaxies and compare to those of other type dwarf galaxies under various assumptions .
First , we assume that LSB dIs are either all oblate or all prolate , and use non-parametric analysis to find the best-fitting distribution of intrinsic shapes .
With this assumption , we find that the scarcity of nearly circular LSB dIs implies , at the 99 % confidence level , that they can not be a population of randomly oriented oblate or prolate objects .
Next , we assume that dIs are triaxial , and use parametric analysis to find permissible distributions of intrinsic shapes .
We find that if the intrinsic axis ratios , \beta and \gamma , are distributed according to a Gaussian with means \beta _ { 0 } and \gamma _ { 0 } and a common standard deviation of \sigma , the best-fitting set of parameters for LSB dIs is ( \beta _ { 0 } , \gamma _ { 0 } , \sigma ) = ( 0.66 , 0.50 , 0.15 ) , and the best fit for BCDs is ( \beta _ { 0 } , \gamma _ { 0 } , \sigma ) = ( 0.66 , 0.55 , 0.16 ) , while the best fit for dEs is ( \beta _ { 0 } , \gamma _ { 0 } , \sigma ) = ( 0.78 , 0.69 , 0.24 ) .
The dIs and BCDs thus have a very similar shape distribution , given this triaxial hypothesis , while the dEs peak at a somewhat more spherical shape .
Our results are consistent with an evolutionary scenario in which the three types of dwarf galaxy have a close relation with each other .