Int_t my_ints;and then to access the ith:-
my_ints[i]Caution: C++ array indeces start from 0!!, so to sum all these integers:-
Int_t sum = 0; for (Int_t index=0; index<10; index++) sum += my_ints[index];
C++ arrays are no more dynamic than FORTRAN ones, so if the application calls for arrays where the size won't be known until execution time then a container object i.e. one that can hold other objects should be used instead, such as ROOT's TObjArray.
Multi-dimensional arrays are possible, although the syntax can drive a FORTRAN programmer a little crazy: each dimension is in a separate pair of  and the order is reversed, so:-
Int_t my_ints;represents 5 arrays, each being an array of 10 integers. It may help to understand C++ nests constructs to develope new ones. So, rather than have multiply dimensioned arrays, C++ reuses the single array concept but allows arrays to be stored in arrays, hence an array of arrays. Then the  closest to the identifier has to be the one that operates first, hence the ordering.
Multi-dimensional arrays are generally very inflexible, C++ won't allow you to play the same games changing array dimensionality when passing the array to a function. This makes writing general purpose matrix processing libraries difficult. Fortunately, C++ also provides a better way of think about matrices - as objects. In this way operations on matrices become member functions of a matrix class.
See operator precedence