First element of vector becoming equal to last element automatically?!?

I am using the following code to solve the laplace equation.

The issue I am facing is that the first element of norm_err which I am setting to 1 is automatically being reset to the last element on each iteration of the while loop. Why is this happening?

clear; clc; close all;

%==================================================
% MAKE n LIST

n_list = [5, 10, 15];

%==================================================
% SOLVE LAPLACE EQUATION FOR EACH n

% PreAllocation
iterations = zeros(1, length(n_list));

for k = 1:length(n_list)
    
    % Keep Track of the iterations
    k
    
    %==================================================
    % MAKE A MATRIX TO HOLD AN n x n GRID
    
    n = n_list(k);
    % actual_grid = zeros(n, n);

    %==================================================
    % INITIALIZE THE GRID
    
    % iterated_grid = zeros(n, n);
    amplitude = 1;
    iterated_grid = amplitude * rand(n, n);
    
    %==================================================
    % SET BOUNDARY CONDITIONS
    
    x0 = 0;
    y0 = 0;
    xf = 1;
    yf = 1;
    
    xx = linspace(x0, xf, n);
    yy = linspace(y0, yf, n);
    
    delta_x = (xf - x0) / (n - 1);
    delta_y = (yf - y0) / (n - 1);
    
    [X, Y] = meshgrid(xx, yy);
    
    actual_grid = X.^2 - Y.^2;
    
    % Set boundary of iterated grid
    iterated_grid(1, :) = actual_grid(1, :);
    iterated_grid(length(yy), :) = actual_grid(length(yy), :);
    iterated_grid(:, 1) = actual_grid(:, 1);
    iterated_grid(:, length(xx)) = actual_grid(:, length(xx));
    
    %==================================================
    % SET AVERAGE VALUE FOR EACH POINT, ITERATE N TIMES
    
    tolerance = 1e-6;
    
    copy_grid = iterated_grid;
    norm_err{k}(1) = 1;
    
    iterations(k) = 0;

    while norm_err{k}(iterations(k) + 1) > tolerance
        
        iterations(k) = iterations(k) + 1;
        
        % Run the iteration through the interior of the matrix
        iterated_grid(2:end-1, 2:end-1) = (0.25) * (iterated_grid(1:end-2, 2:end-1) + iterated_grid(3:end, 2:end-1) + iterated_grid(2:end-1, 1:end-2) + iterated_grid(2:end-1, 3:end) );
        
        difference_grid = iterated_grid - copy_grid;
        norm_err{k}(iterations + 1) = sqrt(sum(sum(difference_grid(2:end-1, 2:end-1).^2)) / ((n-2)*(n-2)));
        copy_grid = iterated_grid;
        
    end
    format long
    disp(norm_err{k}([1, 2, 3, end]));

end

k = 1

   0.000000720425189   0.393537402267598   0.169809678519038   0.000000720425189

k = 
     2

   0.000000950157662   0.423949692990376   0.289627916298907   0.000000950157662

k = 
     3

   1.000000000000000   0.333282811338901   0.163630143300983   0.000000975445388