Run-time error in call to std::transform based on iterator range

I have a simple matrix of 50x40, I fill its elements with scalar values from 0.0f-50.0f as can be seen in code. Then, I truncate the matrix scalar values based on lowerbound and upperbound as per the code. The code works great, no problem.

#include "boost\numeric\ublas\matrix.hpp"
typedef boost::numeric::ublas::matrix<float> matrix;

template <typename T1>
class Truncate{  // Functor used for truncation
public:
    Truncate(T1 lbIn, T1 ubIn) : lb(lbIn), ub(ubIn){}
    float operator()(const float &val) const    {
        if (val <= lb)
            return lb;
        else if (val >= ub)
            return ub;
        else
            return val;
    }
private:
    T1 lb; // lowerBound
    T1 ub; // upperBound
};

void TestMinMax(){
    matrix m1(50, 40);
    for (size_t i = 0; i < m1.size1(); i++) {
        for (size_t j = 0; j < m1.size2(); j++) {
            m1(i, j) = i + j*0.01f;
        }
    }

    Truncate<float> chop(10.4f, 30.55f);
    transform(m1.begin2(), m1.begin2() + m1.size1()*m1.size2(), m1.begin2(), chop); // Works!
    // transform(m1.begin1(), m1.begin1() + m1.size1()*m1.size2(), m1.begin1(), chop); // Gives run-time error

    for (size_t i = 0; i < m1.size1(); i++) {
        for (size_t j = 0; j < m1.size2(); j++) {
            cout << m1(i, j) << "\t";
        }
        cout << endl;
    }
}

But, I would like to draw your attention to a line I have commented out in my code -

// transform(m1.begin1(), m1.begin1() + m1.size1()*m1.size2(), m1.begin1(), chop); // Gives run-time error

If I use this line, instead of the one currently active, I get run-time error. Interestingly, the run-time error is only encountered when the iterator range exceeds 40. What could be the reason for the run-time error?