fitlm with sequential variable names

I'm trying to run multiple regressions using a for loop. This is my code:

(N is the number of columns in the matrices, each column represents one of the 24 hours. Each matrix is a different variable.)

for i = 1:N                    
    table_hourly = table( First_matrix(:,i), Second_matrix(:,i), Third_matrix(:,i),'VariableNames', ...
    {sprintf('Variable1_Hour%d',i) sprintf('Variable2_Hour%d',i) sprintf('Variable3_Hour%d',i)});
    a = sprintf('Variable1_Hour%d',i);
    b = sprintf('Variable2_Hour%d',i);
    c = sprintf('Variable3_Hour%d',i);
    fitlm(table_hourly, 'a ~ b + c', 'Intercept',0,'Robust',1)
end

With this code, I can correctly create a table (table_hourly) for each different hour, with the right column headers. However, when it gets to the regression, fitlm fails to recognise the variable names.

How can I solve this problem?